1 Complete

vii

WATER QUALITY RESEARCH JOURNAL OF CANADA Volume 38, No. 2, 2003 The Application and Suitability of Microbiological Tests for Fecal Bacteria in Pulp Mill Effluents: A Review M.L. TAMPLIN

211–225

An Assessment of Variability of Pulp Mill Wastewater Treatment System Bacterial Communities using Molecular Methods C.J.O. BAKER, R.R. FULTHORPE AND K.A. GILBRIDE

227–242

Innovative Biological Treatment Processes for Wastewater in Canada C.N. MULLIGAN AND B.F. GIBBS

243–265

Binding of Hydrophobic Organic Contaminants to HumaliteDerived Aqueous Humic Products, with Implications for Remediation D.R. VAN STEMPVOORT, S. LESAGE AND H. STEER

267–281

An Assessment of Long-Term Monitoring Data for Constructed Wetlands for Urban Highway Runoff Control A.C. FARRELL AND R.B. SCHECKENBERGER

283–315

Windsor Combined Sewer Overflow Treatability Study with Chemical Coagulation J.G. LI, S. DHANVANTARI, D. AVERILL AND N. BISWAS

317–334

Enhanced Prairie Wetland Effects on Surface Water Quality in Crowfoot Creek, Alberta G.R. ONTKEAN, D.S. CHANASYK, S. RIEMERSMA, D.R. BENNETT AND J.M. BRUNEN

335–359

Physiological and Biochemical Responses of Ontario Slimy Sculpin (Cottus cognatus) to Sediment from the Athabasca Oil Sands Area G.R. TETREAULT, M.E. MCMASTER, D.G. DIXON AND J.L. PARROTT

361–377

Textile Dye Removal by Membrane Technology and Biological Oxidation M. GHOLAMI, S. NASSERI, M.-R. ALIZADEHFARD AND A. MESDAGHINIA

379–391

continued on next page

Removal of Dyes from Aqueous Solutions by Adsorption on Chrome-Tanned Solid Wastes Generated in the Leather Industry S. TAHIRI, A. MESSAOUDI, A. ALBIZANE, M. AZZI, M. BOUHRIA, S.A. YOUNSSI, J. BENNAZHA AND J. MABROUR

393–411

Adsorption of Cadmium Ions onto the Yellow River Sediment S.L. HUANG

413–432

Philip H. Jones Student Award—38th Central Canadian Symposium on Water Quality Research

ix

Water Qual. Res. J. Canada, 2003 Volume 38, No. 2, 221–225 Copyright © 2003, CAWQ

The Application and Suitability of Microbiological Tests for Fecal Bacteria in Pulp Mill Effluents: A Review MARK L. TAMPLIN* Institute of Food and Agricultural Sciences, University of Florida, Gainesville, Florida

Tests for fecal (thermotolerant) and total coliforms applied to drinking and recreational waters are convenient but not necessarily accurate indicators of the presence of pathogens originating from the intestines of warm-blooded animals. These tests may be inappropriate for analyses of complex industrial effluents such as pulp mill effluents, which may contain bacteria that yield a positive result in fecal and total coliform tests, even though no fecal source exists. Consequently, other more specific indicators for the presence of potential pathogens are needed. In cases where the sources of, or validity of, fecal responses are problematic, the direct examination of pathogens, and/or the use of new source-tracking techniques, such as DNA fingerprinting, may be better approaches for protecting public health. Key words: Klebsiella, coliforms, pulp mill, effluent, indicators, E. coli

Introduction Since the early part of the 20th century, water quality has been monitored by methods that include analyses for coliform bacteria, i.e., organisms once believed to be unique to the gastrointestinal tracts of animals (Gerba 1987). It is now recognized that some of the organisms that may be enumerated as total or fecal (thermotolerant) coliforms, e.g., Klebsiella, Citrobacter, and Enterobacter spp., represent a group of organisms that can be found in both animal and non-animal entities, including wood, soils, and vegetation. These non-enteric sources may contribute to elevated responses of fecal coliform assays applied to pulp mill effluents with no sanitary sewer connections (Archibald 2000; Capelenas and Kanarek 1984; NCASI 1971a), even in cases where animal sources of bacteria are low or absent (NCASI 1971b; Clarke et al. 1992). Significant information has appeared in the scientific literature which documents the occurrence and growth of Klebsiella pneumoniae as well as other members of the Enterobacteriacae in natural woods and vegetation (Cosenza et al. 1970; DeGroot and Sachs 1976; Maloy and Robinson 1968). Today, many scientists and policy-makers recognize the ubiquitous nature of coliform bacteria, and since the 1980s there have been important developments in the proposed utilization of more specif-

*

[email protected]

212

TAMPLIN

ic tests for measuring human fecal impact on water. Enterococci and Escherichia coli have been shown to have better predictive values of human gastroenteritis than fecal coliforms in surface, marine and fresh waters (Dufour 1984; Cabelli 1983). However, progress in implementing E. coli and enterococci tests has been slow, and tests for fecal coliforms are still required in many circumstances where bacterial water quality must be established. This includes tests of recreational and shellfish harvesting waters, even though E. coli and enterococci tests would provide at least equal protection with fewer chances for false positives and the associated adverse societal and economic impact. Because of the difficulties associated with developing and utilizing tests for specific waterborne pathogenic bacteria, the desire to employ tests for indicator organisms is reasonable. There are desirable characteristics of indicator organism tests, and these are discussed in the review which follows. It should be noted that the fecal coliform assay applied to pulp mill effluents responds to Klebsiella and E. coli, and that (1) these bacteria multiply in effluent biological treatment systems, (2) this multiplication obviates the proportionality between the fecal coliform analysis result and pathogens which may be present which does not multiply therein, and (3) the survival of the indicator is no longer equal to that of the pathogen. The development of DNA fingerprinting techniques and other advanced methods has provided pathways by which valid assessments of the sources of pathogens tentatively identified via indicator tests may be undertaken. These newer methods may be applied in various situations where it is necessary to differentiate sources of microbial input (Parveen et al. 1999). The review which follows provides an examination of the background, science, issues, and applications concerning the use of indicator bacteria and newer methods to assess water quality, with focus on pulp mill effluents as an example of matrices in which total and fecal coliform assays may be misapplied.

Fecal Bacteria Feces contains a variety of microorganisms that are present in high concentrations of up to 1010 per gram (Fiksdal et al. 1985). By far the dominant intestinal organisms are anaerobic bacteria. Most of these species are not pathogenic to humans, but are essential for maintaining normal functions of the gastrointestinal system. Facultative intestinal bacteria are also mostly non-pathogenic, however this group contains the majority of species that cause human disease, such as pathogenic strains of Campylobacter, Salmonella, E. coli and Shigella. There is justifiable concern regarding microorganisms that originate from humans, because humans serve as reservoirs and reseed the environment with pathogens. We are concerned to a lesser extent with animal feces, because it does not harbor as many human pathogens.

NON-SPECIFIC COLIFORMS IN PULP MILL EFFLUENTS

213

They rarely carry human viruses like Hepatitis A and Norwalk, or bacteria such as Shigella, Salmonella typhi, Salmonella typhimurium, and certain pathogenic strains of E. coli. However, animals can be sources of human pathogens. For example, domestic bovine animals (cows) harbor the parasite Cryptosporidium parvum and E. coli O157:H7; poultry contain Campylobacter spp. and various cold- and warm-blooded animals can carry Salmonella spp.

Indicators of Fecal Contamination Indicators of the Sanitary Quality of Water An indicator is an organism or substance whose presence and concentration signals the occurrence of another entity in a matrix under examination. Addressed by this review is whether fecal coliforms constitute the best indicator for pathogens potentially found in pulp and paper mill effluents. In that it has been acknowledged already that no one indicator may be best in all situations (e.g., U.S. Environmental Protection Agency’s recommendation that enterococci should be substituted for E. coli and fecal coliforms for assessing potential risks associated with recreational marine waters), it is not unreasonable to question other circumstances, e.g., whether fecal coliforms are appropriate indicators for pulp mill effluents. When considering the appropriate indicator for a specific situation, these ideal indicator properties should be considered: •

• • • •

The concentration of the indicator should be higher than that of the hazard it predicts, and the indicator must be detectable when the hazard is present at non-hazardous concentrations. The indicator should not multiply in the environment. The concentration of the indicator should vary proportionally with the concentration of the hazard. The survival of the indicator should be equal to or greater than that of the hazard. The indicator test should be highly reproducible and standardized, so that similar results may be obtained by various laboratories.

Bacterial Water Quality Standards Development The coliform group has been used by the U.S. Public Health Service as an indicator of fecal contamination from as early as 1914 (Gerba 1987). Total coliforms include the aerobic and facultative anaerobic, gram-negative, non-spore-forming, rod-shaped bacteria that ferment lactose with acid and gas production within 48 hours at 35°C (APHA 1989). This group includes E. coli, and various members of the genera Enterobacter, Klebsiella, and Citrobacter. These coliforms are discharged in high numbers (2 x 109 coliforms/day/capita) in human and animal feces (Gerba 1987). However, not all strains of these bacteria are of fecal origin.

214

TAMPLIN

Fecal coliforms are thermotolerant and include all coliforms that can ferment lactose at 44.5°C in 24 hours. This group of organisms includes E. coli and K. pneumoniae, both of which are not specific to human or animal feces. Reports show that these species can be found not only in feces, but also various forms of plant material (Cosenza et al. 1970; DeGroot and Sachs 1976; Knittel et al. 1978; Maloy and Robinson 1968). Members of the fecal streptococci persist well, and have the potential for reproducing in aquatic environments. A subgroup of the streptococci, the enterococci has been suggested as a useful indicator for viruses in seawater (APHA 1992; Bitton 1999). Some have proposed that a ratio of fecal coliforms to fecal streptococci (FC/FS) can serve to discriminate the pollution source as human or non-human origin (Geldreich and Kenner 1969), however others have questioned the usefulness of this approach, showing that the organisms have different survival rates and sensitivities to water treatments (Pourcher et al. 1991). Furthermore, their utility as indicators of human fecal pollution is also criticized in reports showing that enterococci are ubiquitous, and can persist and grow on vegetables and other plant waste (Gauthier and Archibald 2001; Geldreich and Kenner 1969; Mundt 1961, 1963). Fecal streptococci are discussed in greater detail below. Anaerobic bacteria have also been considered as indicators. These include Clostridium perfringens which represents 0.5 percent of the fecal microflora (Fujioka 1997; Payment et al. 1993), Bifidobacteria which are one-third of the fecal microflora (Bitton 1999), and Bacteroides that occur at 1010 cells per gram of feces (Fiksdal et al. 1985). However, these indicators have not gained broad acceptance due to undesirable environmental survival patterns, and/or lack of validation studies. Specific viruses have been suggested as indicators of pathogens, especially viral pathogens. For example, bacteriophages (i.e., viruses of bacteria), are similar to enteric viruses, but are more easily detected and found in higher numbers in wastewater than enteric viruses (Bitton 1980; Goyal et al. 1987). However, it has been questioned whether bacteriophages are good surrogates for enteroviruses in all situations (Gerba 1987). F-specific bacteriophages are not frequently found in human feces but are good indicators of wastewaters and viral contamination in the marine environment (Bitton 1999).

Sources and Ecology of Coliform Bacteria, Considering Wood-Derived Sources E. coli The primary reservoir of E. coli is the intestines of warm-blooded animals. Of the species that constitute the total and fecal coliform groups, E. coli is the most specific to animal and human gastrointestinal tracts. In the majority of studies of pulp mill effluents, E. coli has not been shown to be a dominant component of the flora. In many instances, E. coli is absent


215

when the fecal coliform test is positive and Klebsiella have been detected. Nevertheless, there are reports of E. coli in mill environments, even those with no sanitary sewer connection to process wastewater or effluent treatment (Gauthier and Archibald 2001; Hendry et al. 1982; MeGraw and Farkas 1993; Niemi et al. 1987). For example, MeGraw and Farkas (1993) observed that E. coli was present in the chip screw pressate only during the processing of whole log chips, absent in pressate of two-week-stored saw mill chips, and they postulated that E. coli grew in log wash water. Furthermore, they found E. coli in fresh whole-log and sawmill chips from poplar, maple, and basswood/poplar mix, but not in the bark of these trees, indicating root transfer rather than surface contamination. This finding is similar to that of Bagley et al. (1978), who isolated Klebsiella and other coliform bacteria from inside trees. More recently, Gauthier and Archibald (2001) analyzed numerous samples from seven Canadian pulp and paper mill water systems and observed that E. coli replicated within the mill environment. This occurred even though all but one of the mills had no sewer input, and most mills disinfected their input water. The primary clarifier was found to be a major site of E. coli (and other members of the coliform group) replication. Whole dry logs stored for up to one year did not yield E. coli, likely due to the lethal effects of dessication (MeGraw and Farkas 1993). In another study, Niemi et al. (1987) reported that E. coli multiplied and persisted as the dominant thermotolerant coliform in a pulp and board mill that used birch as the raw material and ammonium sulfate as the process chemical. E. coli has also been shown to persist in tropical soils and may not be an appropriate indicator of fecal pollution in these environments (Byappanahalli and Fujioka 1998). These findings indicate that the presence or absence of E. coli in mill process waters may be specific to some operations, or to local circumstances. In this regard, in cases where sewage is known to enter the system, the proportion of coliforms from sewage and the proportion of wood-derived coliforms simply growing there will not be known. This can markedly complicate efforts to link measured coliform and enterococci levels to levels of sewage contamination or health hazards. Klebsiella Coliform bacteria have been isolated from pristine sites, including those of tropical rainforests (Rivera et al. 1988). The non-specific nature of coliform bacteria in living and cut wood (e.g., tree needles, bark) and other botanical environments has been established in many studies (Cosenza et al. 1970; DeGroot and Sachs 1976; Knittel et al. 1978; Maloy and Robinson 1968). In most cases, the coliforms identified have been Klebsiella spp., Enterobacter spp., and Citrobacter spp. Hypotheses about the origins of these organisms include entrance through the root systems and up the trunks, as evidenced by scanning electron microscopy of tracheal tubes (Bagley et al. 1978). As reported by Bagley et al. (1978), the high numbers of Klebsiella spp. and Enterobacter spp. in samples of wood “clear-

216

TAMPLIN

ly indicate that these organisms are indigenous to the wood.” It was observed that Klebsiella were capable of multiplying to very high numbers (106/mL) in aqueous extracts of sawdust and on the surfaces of vegetables (103/g of surface peel). In addition, Gauthier and Archibald (2001) found that K. pneumoniae, K. cloacae and E. coli grew to very high densities in sterilized raw, combined mill effluent. The growth of these species in relatively nitrogen-limited environments, such as those of pulp mill effluents, likely results from their ability to fix atmospheric nitrogen and to grow at mesophilic temperatures on simple sugars (LeChevallier et al. 1996; Neilson and Sparell 1976). For example, Knowles et al. (1974) observed that 32% of 129 Klebsiella isolates from pulp mills, lakes, rivers, and drainage and sewage systems possessed nitrogen-fixing ability. In addition, Gauthier et al. (2000) demonstrated active in situ nitrogen fixation by coliforms in a pulp mill primary clarifier. However, many modern pulp mill effluent treatment systems operate using added nutrients, and the nitrogen limitation observed in untreated effluents may not be the only factor in promoting Klebsiella proliferation. Fecal Streptococci and Enterococci Fecal streptococci have been used for many years as indicators of fecal pollution (Bartley and Slanetz 1960), and more recently have gained greater acceptance as an indicator for fresh and marine surface bathing waters (Dufour 1984; Cabelli 1983). They appear to show some properties of a valid indicator for enteroviruses, since (1) their survival in wastewater treatment systems is similar to that of viruses, (2) they are not harmful, and (3) they do not multiply in water (Bartley and Slanetz 1960). However, the latter property may not apply to all waters, since other investigators present evidence that in six pulp and paper mill water systems having no sewage inputs, enterococci were numerous and always present (Gauthier and Archibald 2001). The fecal streptococci group contains a number of Streptococcus species, including S. faecalis, S. faecium, S. avium, S. bovis, S. equinus, and S. gallinarum, and their primary habitat is the gastrointestinal tract of warm-blooded animals (APHA 1992). At one time it was thought that the presence of certain fecal streptococci species could be used to differentiate the source of fecal pollution, however studies have shown less host specificity, and this application is not recommended. Furthermore, several of these species have been reported from non-mammalian sources, such as insects and plants (Clausen et al. 1977; Geldreich et al. 1964). The ratio of fecal coliforms to fecal streptococci is also not recommended for differentiating fecal sources (APHA 1992). The enterococci are a subgroup of the fecal streptococci, and include the species S. faecalis, S. faecium, S. gallinarum, and S. avium. They can be isolated from samples using the procedures for fecal streptococci, but the enterococci have the ability to grow in 6.5% sodium chloride, at pH 9.6,


217

and at 10°C and 45°C. Studies of swimming-associated gastroenteritis at fresh and marine bathing beaches, showed that enterococci are more effective indicators of sanitary water quality than fecal coliforms (Cabelli 1983; Dufour 1984). The recommended guideline for enterococci is 33/100 mL for fresh waters, and 35/100 mL for marine waters, based on the geometric mean of at least five samples per 30-d period of the swimming season (U.S. Environmental Protection Agency 1986).

Public Health Risk Potential of Waterborne/Airborne Coliform Bacteria Klebsiella Since Klebsiella are known to exist in pulp mill environments at elevated concentrations, some researchers have questioned whether this species constitutes a potential health hazard. The consensus of several published studies is that waterborne exposure to Klebsiella in general does not cause adverse health effects, likely because most are normal flora of humans and are easily controlled by host defense mechanisms. Furthermore, there is no known epidemiological evidence that the presence of Klebsiella in recreational waters constitutes a public health hazard (Duncan 1988a,b; NCASI 1972). Klebsiella infections are opportunistic in nature, occurring in immunocompromised persons such as alcoholics, and are normally acquired in hospital and other health care settings. Although a few reports suggest that there is potential concern for occupational exposure, this assertion is not substantiated by epidemiological reports (Duncan 1988a; Kanarek and Caplenas 1981; Niemelä et al. 1985). The Ontario Ministry of the Environment commissioned a review of Klebsiella to address concerns about its prevalence in pulp and paper mills. The resulting paper by Duncan (1988a) concluded that the organism is an opportunistic pathogen, and rarely causes community-based infections. That report further states that pneumonia caused by K. pneumoniae normally occurs among hospitalized individuals who are immunocompromised. In otherwise healthy persons, it can cause urinary tract infections transmitted by the body’s own Klebsiella flora, although other intestinal flora are more likely to be etiological agents of urinary tract infections. Duncan (1988a,b) concluded that recreational waters are not associated with Klebsiella infection. Earlier studies by Kanarek and Caplenas (1981) found that 4% to 97% of fecal coliforms in pulp and paper mill waters were Klebsiella spp., and that similar biotypes, as expected, were found in nasal cavities of workers. The authors found no conclusive evidence that Klebsiella colonization significantly differed between exposed and unexposed workers. Unfortunately, at the time of these studies, DNA fingerprinting techniques were not widely available; if so, they could have confirmed whether the isolates were normal flora of the workers or those from the mill environment.

218

TAMPLIN

In a similar study, Niemelä et al. (1985) reported the incidence of selected microbes in the proximal nasal cavity of paper industry workers, and showed that a variety of organisms could be isolated from both control (personnel in dry work areas) and test groups (personnel in wet work areas). Significant differences between the groups were reported for gramnegative rods, yeast, molds, and Klebsiella species, and presumably reflected the higher incidence of aerosols in the wetter environments. Isolation of these bacteria from nasal cavities would be expected. These same investigators reported a “lack of association of nasopharyngeal symptoms with either exposure to aerosols or nasal microbial contamination.” Other studies have found few phenotypic differences among naturally occurring isolates and those from healthy humans. For example, researchers have shown that isolates of Klebsiella from natural receiving waters differ only in the production of indole from Klebsiella isolated from non-hospitalized humans (NCASI 1975). Other characteristics, including serotype diversity and low antibiotic resistance, were similar among these Klebsiella strains (Matsen et al. 1974; NCASI 1975). Similar to the case for pulp mill effluents, Klebsiella can also represent the predominant species of elevated fecal coliforms in shellfish. U.S. Food and Drug Administration researchers investigated the virulence of these isolates in a mouse model and found that the majority of strains were non-pathogenic (Boutin et al. 1986). Escherichia coli The majority of E. coli strains are non-pathogenic, and can be found in high concentrations in the intestinal tract of most warm-blooded animals. In some instances, E. coli can be pathogenic and are classified as enterotoxigenic, enteropathogenic, enterohemorrhagic, enteroinvasive, and enteroaggregative (Guerrant and Thielman 1995; Levine 1987). In a study of E. coli isolates from multiple paper and pulp mill waters, none of the isolates possessed virulence properties associated with pathogenic strains (Gauthier and Archibald 2001). This information indicates that E. coli isolates from mill effluents with no obvious fecal input may be tested for virulence properties to assess human health risks.

Association of Pulp Mill Effluents with Coliform Bacteria Dating back to 1960, the National Council for Air and Stream Improvement (NCASI) conducted extensive studies investigating the occurrence of the coliform group of bacteria in process and natural waters, the growth of Klebsiella in process waters, as well as more suitable indicator organisms. In a two-year study on the Rainy River located between Ontario and Minnesota (NCASI 1971a), over 2000 presumptive total coliforms were isolated, of which approximately 30% and 15% were fecal coliforms and E. coli, respectively. In a separate study of 12 mills, NCASI (1971a) concluded that process areas and streams were very conducive to proliferation of


219

Klebsiella-like organisms, even when sanitary and process waste streams were separated. Another study conducted from 1966 to 1967 showed that fecal coliforms, fecal streptococci, and enteric pathogens were commonly isolated from remote sampling stations with low human use (NCASI 1971b). Various other studies have determined that E. coli and enterococci are better indicators of fecal contamination. A study of pulp mill effluents from six northwestern mills, conducted by NCASI, found that E. coli represented 0 to 40% of the fecal coliform population. Also in this study, when two of the six sites had fecal coliform levels above the standard, none violated standards when E. coli was enumerated, and levels ranged from none detected to seven per 100 mL. In an investigation of Wisconsin pulp and paper mills for fecal coliforms and Klebsiella, up to 90% of non-fecal source thermotolerant K. pneumoniae were falsely identified as fecal source bacteria (Caplenas and Kanarek 1984). Due to a lack of specificity in the fecal coliform test, the authors recommended that a more reliable health risk assessment of fecal bacteria be used. Caplenas et al. (1981) investigated three pulp and paper mills for fecal coliforms and K. pneumoniae bacterial concentrations. They reported that input freshwater streams contained less than 10 cells per 100 mL, yet elevated concentrations could be traced from the early pulping stages to water processing reuse systems. Concentrations of both thermotolerant and non-thermotolerant strains ranged from 4 x 104 to an estimated 3 x 106 per 100 mL. Wastewater treatment had limited effects on reducing K. pneumoniae concentrations. Similar observations have been reported for seven Canadian pulp and paper mill water systems in which very high densities of Klebsiella spp. were found in process waters and biosolids, even though only one had a sewage connection and most input waters were disinfected (Gauthier and Archibald 2001). Richards (1995) provides a thorough evaluation of, and potential solution to, fecal coliform permit violations at a pulp and paper mill in Virginia, where process waters were connected to sanitary sewers. In exhaustive studies of the potential causes for elevated effluent fecal coliforms, Richards examined both sanitary and process wastewater streams, and concluded that the primary problem was Klebsiella growth in process waters. The author’s recommendations included changing to a better indicator system, such as enterococci or E. coli and, although not supported by the observations, removing the connection between the sanitary discharge and the process waste system.

Common Test Formats for Measuring and Identifying Indicator Bacteria Most Probable Number The most probable number (MPN) technique is a simple procedure for estimating the number of bacteria in a sample, and is widely used

220

TAMPLIN

for water and shellfish-harvesting waters. The method is simple to perform, and typically has greater sensitivity than membrane filter or plating methods, because the organism is not subjected to dessication on the surface of a medium, nor changes in pressure caused by filtration. The main drawbacks to the technique are the labour in preparing and conducting the tests, and the wide variation in MPN values which may result when few sample replicates are tested. However, the latter situation can be improved by increasing the number of replicates, such as by using a 10-tube or 96-well device, versus the more common three- and five-tube formats. Membrane Filtration Membrane filtration (MF) was developed to improve the accuracy of, and reduce the time and cost for, enumerating coliforms, fecal coliforms, E. coli, enterococci, and other organisms (Dufour and Cabelli 1975). Unlike the MPN broth method, MF provides isolated colonies that can be enumerated on a gridded surface. Current test formats normally incorporate selective and differential components in an agar medium, onto which the filter is placed and incubated. The MF formats for E. coli and enterococci are likely to move more laboratories away from the less fecal-specific fecal coliform MPN test. Even with these advantages, there are some limitations that should be noted. Samples with high turbidity are not recommended for this procedure, since particulate matter and high background flora cause interference. Also, samples with high levels of target organism and interfering bacteria, such as in sewage, need to be diluted appropriately. Another consideration is that a representative number of colonies must be selected from the filter when extrapolating the confirmed species to other colonies with similar morphology. Otherwise, there will be a gross over- or underestimation. Both MF and MPN broth tests can be used to estimate levels of E. coli and enterococci (APHA 1992), although the MF procedure is not normally recommended for turbid samples with low bacterial counts. For enterococci, the MF technique involves filtration of the sample, incubation on agar, followed by incubation of the same membrane on a substrate medium (APHA 1992). The MPN procedure for enterococci involves culture in broth, followed by culture on agar (APHA 1992). MUG Test The MUG test for E. coli has gained widespread acceptance due to its simplicity and rapid format (APHA 1992). The majority of E. coli strains are unique in producing the enzyme β-glucuronidase, which can hydrolyze 4-methylumbelliferyl-beta-D-glucuronide (MUG) and produce a fluorescent compound under a 366-nm ultraviolet light. MUG, as well as analogs that can be visually perceived, and have been conveniently incorporated into broths, agar, and other rapid test formats.


221

New Techniques for Tracing Sources of Microorganisms Certain questions that are historically raised concerning sources of fecal bacteria can now be addressed with newer methods that permit microorganisms to be identified and tracked (DNA “fingerprinted”). Using these techniques, a bacterial strain may be traced from its source to different points throughout a production environment. Such techniques can determine whether, for example, E. coli strains originate from human or animal sources. Pulsed-Field Gel Electrophoresis There is extensive literature showing that pulsed-field gel electrophoresis (PFGE) is a useful tool for discriminating bacterial strains (Bohm and Karch 1992; Buchrieser et al. 1991; Buchrieser et al. 1995; Prevost et al. 1992; Schoonmaker et al. 1992). The principle of PFGE is based on the distance that large DNA fragments migrate when subjected to an electrical field. It has been used extensively in epidemiological studies, but relatively less in environmental research. PFGE would very useful when tracing the movement of a specific bacterial strain from process system input through discharge. Ribotyping A more conservative DNA fingerprinting method is ribotyping, whereby the size of DNA fragments that code for 16s and 23s ribosomal RNA genes are measured. The technique is less discriminatory than PFGE, but can be used to determine the type of input source (Prevost et al. 1992; Tamplin et al. 1996). For example, certain ribotypes have been associated with specific bacterial species and subtypes (Prevost et al. 1992; Snipes et al. 1992; Tamplin et al. 1996). Parveen et al. (1999) showed that ribotyping can differentiate sources of E. coli as either human or nonhuman, and they have applied the technique to solving pollution problems for certain municipalities. Multiple Antibiotic Resistance An organism’s sensitivity to antibiotics can also be used to evaluate its source. For example, human E. coli are much more resistant to multiple antibiotics than non-human (e.g., wildlife) isolates (Parveen et al. 1998). The technique is rapid and inexpensive, and most suitable when human sources of fecal pollution are suspected.

Conclusions In general, assessing the safety of water will always be a dynamic process, involving the recognition of new hazards, establishing new critical limits, and developing more accurate and sensitive methods. It should

222

TAMPLIN

not be surprising, therefore, that coliform tests have become increasingly out-dated with the advent of better indicators and more sophisticated exposure assessment techniques. Specific to pulp mill effluents, the following conclusions are important, based upon the review presented here: •

•

•

Pulp mill effluents may yield positive responses when analyzed for coliform bacteria and for fecal coliform bacteria. These responses are usually due to the presence of members of the Klebsiella species in the effluents. These bacteria are not pathogenic, and may not indicate the presence of pathogens. In some cases, the positive responses of pulp mill effluents to coliform and fecal coliform assays may be due to the presence of one or more strains of E. coli. In certain instances, these strains may be able to reproduce, however these strains are normally not of human gastrointestinal origin (except possibly in cases where there is a sanitary sewer hookup), and are brought into the mills with the furnish, i.e., they are incorporated into the tree components in the forest. The strains of E. coli thus encountered are not anticipated to be pathogenic. Comprehensive research to trace the origins of E. coli in mill environments, such as using DNA fingerprinting, has not been conducted. The use of coliform and fecal coliform tests to assess the bacteriological quality of pulp mill effluents does not yield information useful for assessing the potential for exposure to disease-causing organisms. More specific indicators, or direct tests, of human pathogens are needed to better estimate human health risk.

References American Public Health Association (APHA). 1989. Standard methods for the examination of water and wastewater, 17th edition. American Public Health Association, Washington, DC. American Public Health Association (APHA). 1992. Standard methods for the examination of water and wastewater, 18th edition. American Public Health Association, Washington, DC. Archibald F. 2000. The presence of coliform bacteria in Canadian pulp and paper mill water systems - a cause for concern? Water Qual. Res. J. Canada 35:1–22. Bagley ST, Seidler RJ, Talbot HW, Morrow JE. 1978. Isolation of Klebsiella from within living wood. Appl. Environ. Microbiol. 36:178–185. Bartley CH, Slanetz LW. 1960. Types and sanitary significance of streptococci isolated from faeces, sewage and water. Am. J. Pub. Health 50:1545–1552. Bitton G. 1980. Introduction to environmental virology. John Wiley & Sons, New York. Bitton G. 1999. Wastewater microbiology, 2nd ed. John Wiley & Sons, New York. Bohm H, Karch H. 1992. DNA fingerprinting of Escherichia coli O157:H7 by pulsed-field electrophoresis. J. Clin. Microbiol. 30:2169–2172. Boutin BK, Spaulding PL, Twedt RM. 1986. Evaluation of the enteropathogenic-


223

ity of Klebsiella pneumoniae isolates from summer-harvested Louisiana oysters. J. Food Prot. 49:442–444. Buchrieser C, Brosch R, Rocourt J. 1991. Use of pulsed-field gel electrophoresis to compare large DNA-restriction fragments of Listeria monocytogenes strains belonging to sero-groups 1/2 and 3. Int. Syst. Microbiol. 14:297–304. Buchresier C, Gangar VV, Murphree RL, Tamplin ML, Kaspar CW. 1995. Multiple Vibrio vulnificus clones in oysters as demonstrated by clamped homogenous electric fields (CHEF) gel electrophoresis. Appl. Environ. Microbiol. 61:1163–1168. Byappanahalli MN, Fujioka RS. 1998. Evidence that tropical soil environments can support the growth of Escherichia coli. Water Sci. Tech. 38:171–174. Cabelli VJ. 1983. Health effects criteria for marine waters. EPA-600/1-80-031, US Environmental Protection Agency, Cincinnati, Ohio. Caplenas NR, Kanarek MS. 1984. Thermotolerant non-fecal source Klebsiella pneumoniae: validity of the fecal coliform test in recreational waters. Am. J. Public Health 74:1273–1275. Caplenas NR, Kanarek MS, Dufour AP. 1981. Source and extent of Klebsiella pneumoniae in the paper industry. Appl. Environ. Microbiol. 42:779–785. Clarke T, Mitchell C, Donnison A. 1992. Bacteriological water quality of pulp and paper mill effluents: the problem of Klebsiella pneumoniae, p. 171–181. In Proceedings of the 1992 TAPPI International Environmental Conference. Clausen EM, Green BL, Litsky W. 1977. Fecal streptococci: indicators of pollution. Bacterial indicators, p. 247–264. In Hoadley AW and Dutka BJ (ed.), Health hazards associated with water. ASTM STP 635, American Society for Testing and Materials. Cosenza BJ, McCreary M, Buck JD, Shigo AL. 1970. Bacteria associated with discolored and decayed tissues in beech, birch, and maple. Phytopathology 60:1547–151. DeGroot RC, Sachs IB. 1976. Permeability, enzyme activity, and pit membrane structure of stored southern pines. Wood Sci. 9:89–96. Dufour AP. 1984. Health effects criteria for fresh recreational waters. EPA-600/184-004, US Environmental Protection Agency, Cincinnati, Ohio. Dufour AP, Cabelli VJ. 1975. Membrane filter procedure for enumerating the component genera of the coliform group in seawater. Appl. Microbiol. 29:826–833. Duncan IBR. 1988a. Health significance of Klebsiella in the environment. Ontario Ministry of the Environment Publication. March. Duncan IBR. 1988b. Waterborne Klebsiella and human disease. Toxicity Assessment 3:581–598. Fiksdal L, Maki JS, LaCroix SJ, Staley JT. 1985. Survival and detection of Bacteroides spp., prospective indicator bacteria. Appl. Environ. Microbiol. 49:148–150. Fujioka RS. 1997. Indicators of marine recreational water quality, p. 176–183. In Hurst CJ, Knudsen GR, McInerney MJ, Stezenbach LD, Walters MV (ed.), Manual of Environmental Microbiology. ASM Press, Washington, DC. Gauthier F, Archibald F. 2001. The ecology of “fecal indicator” bacteria commonly found in pulp and paper mill water systems. Water Res. 35:2207–2218. Gauthier F, Neufeld JD, Driscoll BT, Archibald FS. 2000. Coliform bacteria and nitrogen fixation in pulp and paper mill effluent treatment systems. App. Environ. Microbiol. 66:5155–5160. Geldreich EE, Kenner BA. 1969. Concepts of fecal streptococci in stream pollution. J. Water Pollut. Control Fed. 41:R336–R341.

224

TAMPLIN

Geldreich EE, Kenner BA, Kabler PW. 1964. The occurrence of coliforms, fecal coliforms, and streptococci on vegetation and insects. Appl. Environ. Microbiol. 21:63–69. Gerba CP. 1987. Phage as indicators of fecal pollution, p. 197–209. In Goyal SM, Gerba CP, Bitton G (ed.), Phage ecology. Wiley-Interscience, New York. Goyal SM, Gerba CP, Bitton G. 1987. Phage ecology. Wiley-Interscience, New York. Guerrant RL, Thielman NM. 1995. Types of Escherichia coli enteropathogens, p. 687–690. In Blaser MJ, Smith PD, Ravdin JI, Greenberg HB, Guerrant RL (ed.), Infection of the gastrointestinal tract. Raven Press, New York. Hendry GS, Janhurst, Horsnell G. 1982. Some effects of pulp and paper wastewater on microbiological water quality of a river. Water Res. 16:1291–1295. Kanarek MS, Caplenas NR. 1981. Epidemiological study of Klebsiella pneumoniae among pulp and paper mill workers. U.S. Environmental Protection Agency, EPA-600/S181-023. Knittel MD, Seidler RJ, Eby C, Cabe LM. 1978. Colonization of the botanical environment by Klebsiella isolates of pathogenic origin. Appl. Environ. Microbiol. 34:557–563. Knowles R, Neufeld R, Simpson S. 1974. Acetylene reduction (nitrogen fixation) by pulp and paper mill effluents and by Klebsiella isolated from effluents and environmental situations. Appl. Microbiol. 28:608–613. LeChevallier MW, Welch NJ, Smith DB. 1996. Full-scale studies of factors related to coliform regrowth in drinking water. Appl. Environ. Microbiol. 62: 2201–2207. Levine MM. 1987. Escherichia coli that cause diarrhea: enterotoxigenic, enteropathogenic, enteroinvasive, enterohemorrhagic, and enteroadherent. J. Infect. Dis. 155:377–389. Maloy OC, Robinson BS. 1968. Microorganisms associated with heart rot in young grand fir. Can. J. Bot. 46:306–309. Matsen JM, Spindler JA, Blosser RO. 1974. Characterization of Klebsiella isolates from natural receiving waters and comparison with human isolates. Appl. Microbiol. 28:672–678. MeGraw SR, Farkas MO. 1993. Escherichia coli: a potential source of native fecal coliforms in pulp and paper mill effluents. Pulp and Paper Canada 94:158–160. Mundt JO. 1961. Occurrence of enterococci: bid, blossom, and soil studies. App. Microbiol. 9:541–546. Mundt JO. 1963. Occurrence of enterococci on plants in a wild environment. App. Microbiol. 11:141–144. NCASI. 1971a. Experience with indicator organism tests in determining the bacteriological quality of pulp and papermill effluents and their receiving waters. National Council for Air and Stream Improvement, Inc., Research Triangle Park, NC, Tech. Bull. 244. NCASI. 1971b. Recent field studies of sanitary water quality in receiving waters. National Council for Air and Stream Improvement, Inc., Research Triangle Park, NC, Tech. Bull. 246. NCASI. 1972. Klebsiella pneumoniae infection: a review with reference to the waterborne epidemiologic significance of K. pneumoniae presence in the natural environment. National Council for Air and Stream Improvement, Inc., Research Triangle Park, NC, Tech. Bull. 254.


225

NCASI. 1975. Further studies of the sanitary significance of Klebsiella pneumoniae occurrence in mill effluents and surface waters. National Council for Air and Stream Improvement, Inc., Research Triangle Park, NC, Tech. Bull. 279. Neilson AH, Sparell L. 1976. Acetylene reduction (nitrogen fixation) by Enterobacteriacae isolated from paper mill process waters. Appl. Environ. Microbiol. 32:197–205. Niemelä SI, Vaatanen P, Mentu J, Jokinen A, Jappinen P, Sillanpaa P. 1985. Microbial incidence in upper respiratory tracts of workers in the paper industry. Appl. Environ. Microbiol. 50:163–168. Niemi RM, Niemela S, Mentu J, Siitonen A. 1987. Growth of Escherichia coli in a pulp and cardboard mill. Can. J. Microbiol. 33:541–545. Parveen S, Murphree RL, Edmiston L, Kaspar CW, Portier KM, Tamplin ML. 1998. Association of multiple antibiotic resistance profiles with point and non-point sources of E. coli in Apalachicola Bay. Appl. Environ. Microbiol. 63:2607–2612. Parveen S, Edmiston L, Portier KM, Tamplin ML. 1999. Association of ribotype profile with human and non-human E. coli in Apalachicola Bay. Appl. Environ. Microbiol. 65:3142–3147. Payment P, Franco E, Ssiemiatycki J. 1993. Absence of relationship between health effects due to tapwater consumption and drinking water quality parameters. Water Sci. Technol. 27:137–143. Pourcher AM, Devriese LA, Hernandez JF, Delattre JM. 1991. Enumeration by a miniaturized method of E. coli, Streptococcus bovis and enterococci as indicators of the origin of faecal pollution of waters. J. Appl. Bacteriol. 70:525–530. Prevost G, Jaulhac B, Piemont Y. 1992. DNA fingerprinting by pulsed–field gel electrophoresis is more effective than ribotyping in distinguishing among methicillin-resistant Staphylococcus aureus isolates. J. Clin. Microbiol. 30:967–973. Richards AC. 1995. Klebsiella interference in fecal coliform testing, p. 11–19. In Proceedings of the 1995 TAPPI International Environmental Conference. Rivera SC, Hanzen TC, Toransos GA. 1988. Isolation of fecal coliforms from pristine sites in a tropical rain forest. Appl. Environ. Microbiol. 54:512–517. Schoonmaker D, Heimberger T, Birkhead G. 1992. Comparison of ribotyping and restriction enzyme using pulsed-field gel electrophoresis for distinguishing Legionella pneumophilia isolates obtained during a nosocomial outbreak. J. Clin. Microbiol. 30:1491–1498. Snipes KP, Wild MA, Miller MW, Jessup DA, Silflow RL, Foreyt WJ, Carpenter TE. 1992. Using ribosomal RNA gene restriction patterns in distinguishing isolates of Pasteurella haemolytica from bighorn sheep (Ovis canadensis). J. Wildl. Dis. 28:347–354. Tamplin ML, Jackson JK, Buchreiser C, Murphree RL, Portier KM, Gangar V, Miller LG, Kaspar CW. 1996. Pulsed-field gel electrophoresis and ribotype profiles of clinical and environmental Vibrio vulnificus isolates. Appl. Environ. Microbiol. 62:3572–3580. U.S. Environmental Protection Agency. 1986. Ambient water quality criteria for bacteria – 1986. EPA-440/5-84-002, U.S. Environmental Protection Agency, Cincinnati, Ohio.


An Assessment of Variability of Pulp Mill Wastewater Treatment System Bacterial Communities using Molecular Methods CHRISTOPHER J.O. BAKER,1 ROBERTA R. FULTHORPE2 AND KIMBERLEY A. GILBRIDE3* 1Ecopia

BioSciences Inc., Montreal, Quebec of Physical Sciences, University of Toronto at Scarborough, 1265 Military Trail, Scarborough, Ontario M1C 1A4 3Department of Chemistry and Biology, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3 2Division

The DNA fingerprinting techniques, 16S-restriction fragment length polymorphism (16S-RFLP), ribosomal intergenic spacer analysis (RISA) and repetitive extragenic palindrome PCR (Rep-PCR), were used for analyzing the bacterial communities of seven pulp and paper wastewater treatment systems. All three methods generate DNA fingerprints that can be compared using the computerassisted program, Gelcompar©. Community similarity coefficients were based on quantitative determinations of both the positions of the DNA bands and the band intensities in order to compare the relative differences in the populations. Unique 16S-RFLP DNA fingerprints were observed for each mill suggesting that individual mills contained phylogenetically different communities. However this method was not sensitive enough to detect differences within a mill treatment system from different locations or from different sampling times. The RISA method, which generated more complex fingerprints than 16S-RFLP, could, for some mills, discern differences between samples. The Rep-PCR technique, however, showed the highest degree of resolution and produced not only distinct patterns for each mill but also distinct fingerprints for the temporal and spatial samples from some of the treatment systems. The sensitivity of this method might potentially be used to monitor the stability of the bacterial community within a secondary treatment system. Key words: 16S-RFLP, RISA, Rep-PCR, diversity, secondary treatment, pulp mills

Introduction To treat pulp mill effluents, secondary wastewater treatment systems have been used for decades for the reduction of BOD (biochemical oxygen demand), COD (chemical oxygen demand), AOX (adsorbable organic halogens), and resin acids in the final effluent (Lindstrom and Mohamed 1988; Bryant and Berkley 1991; Kostyal et al. 1997). The two main types of biotreatment processes employed by the pulp and paper industry are the

* Corresponding author; [email protected]

228

BAKER ET AL.

aerated lagoon (aerated stabilization basin) and the activated sludge system. Regardless of the design, both these secondary wastewater treatment systems are engineered systems that employ a complex community of microorganisms to process vast quantities of wastewater everyday, and to degrade or transform toxic and complex compounds into simple and harmless products that can be discharged into receiving waters. There are relatively few studies of the microbial communities within pulp and paper treatment systems. In 1992, Liss and Allen (1992) surveyed three Northern Ontario mills and showed that the culturable community was dominated by aerobic and facultative anaerobic heterotrophs. Fulthorpe et al. (1993) showed that the culturable heterotrophic bacteria from one of the mills were dominated by populations of Ancylobacter aquaticus, Pseudomonas, and a methylotrophic group (later identified as Xanthobacter). Mohn et al. (1999) isolated several bacterial species capable of resin acid degradation from pulp mill treatment systems including Sphingomonas, Zoogloea and Pseudomonas. The presence and importance of these bacterial groups in all mill treatment systems however, is questionable since culturable isolates have been found to comprise less than 10% of the bacterial population in complex communities and environmental samples (Ward et al. 1992; Wagner et al. 1993; Amann et al. 1995; Torsvik et al. 1996, 1998). Therefore, community level approaches that take into account both the culturable and non-culturable microbes in the treatment system are important in the analysis of the community structure and composition. Phenotypic analysis on whole communities has been carried out using carbon substrate utilization profiles. Victorio et al. (1996) and Schneider et al. (1997) demonstrated the utility of GN and MT Biolog plates for investigating various wastewater samples. They were able to generate phenotypic fingerprints of communities and demonstrate their superiority over traditional microbial assays for identifying shifts in and comparing mill biotreatment microbial communities. However such growth-dependent phenotypic fingerprints do not seem to reflect the in situ functional capabilities nor the identity of the community (Smalla et al. 1998). Genotypic analysis methods are invaluable for the identification and classification of microbes. They are independent of culture techniques, do not rely on gene expression and can be applied to the whole community. Each of these methods permits a certain level of phylogenetic resolution from genera to species to strain when used to look at pure cultures. Although these methods were developed with pure cultures they have been used to monitor phylogenetic changes in simple communities (Massol-Deya et al. 1997; Matheson et al. 1997) and have the potential to be useful in characterizing complex microbial communities (Grey and Herwig 1996; Smit et al. 1997; Yu and Mohn 2001). Furthermore, quantitative comparisons of microbial communities using molecular methods can be analyzed using computer-assisted software packages that can quantify differences in DNA fingerprints (Rademaker and de Bruijn 1997; Rademaker et al. 1999). The software is capable of either basing similarities on the absence or presence of com-

PULP MILL WASTEWATER TREATMENT SYSTEM COMMUNITIES

229

mon DNA bands in the profiles or basing similarities on both band position and intensity. Similarities calculated based on band position alone tend to overestimate the similarity between populations and therefore to compare relative similarities between communities, band intensity should be included in the analysis (Yang and Crowley 2000; Yu and Mohn 2001). In this paper, three molecular methods based on polymerase chain reaction (PCR) amplification of target sites were used. The first method involved the amplification of the 16S rRNA gene followed by restriction analysis (16S-RFLP; restriction fragment length polymorphism) (Moyer et al. 1996). The second method involved the amplification of the intergenic spacer region between the 16S and 23S rRNA genes (RISA) (ribosomal intergenic spacer analysis) (Borneman and Triplett 1997). The last method involved the amplification of repetitive extragenic palindromic sequences (Rep-PCR) (de Bruijn et al. 1996; Fulthorpe et al. 1998; Versalovic et al. 1991). The first two methods are based on DNA fingerprints generated from a single gene or region of the genome while the third technique generates a DNA fingerprint based on sequences throughout the whole genome. The DNA patterns that are generated were compared using computer-assisted cluster analysis of both band position and intensity to estimate the genotypic relatedness of the profiles. Although a few studies have reported on the bacterial community structure in individual aerated lagoon or activated sludge systems used in the pulp and paper industry (Yu and Mohn 2001; Muttray et al. 2001), no studies exist that have directly compared the bacterial community composition of several systems to each other. The aim of this study was to characterize the bacterial communities from seven pulp and paper mill secondary wastewater treatment systems using DNA-based methods to determine how genotypically similar they were. Furthermore, we examined whether those differences could be related to differences in wood furnish (the type of wood used for pulping) or the type of effluent treatment process (aerated lagoon versus activated sludge). We also compared spatial and temporal bacterial community profiles within several of the mill biotreatment systems.

Experimental Procedures Sample Collection More than eighty treatment system effluent samples were received from seven bleach kraft pulp and paper mills from Brazil, New Zealand, U.S.A. and Canada (Table 1). Mills were assigned codes based on the wood furnish, S for softwood, H for hardwood and M for mills using a mixture of both hardwood and softwood. The temporal samples were collected approximately one-third of the way through the treatment system at all mills. Spatial samples were collected for analysis at additional sites in three mills (H1, H2 and M1). All samples were grab samples (1 litre) collected by mill personnel. The samples were shipped on ice and arrived in the lab

6 8 4 3 3 3 4

No. of times sampled 1 1 6 2 3 1 1

No. of locations sampled Pine Radiata pine Eucalyptus Poplar/aspen Mixed hard and soft Mixed hard and soft Mixed hard and soft

Wood furnish

S for softwood furnish, H for hardwood furnish and M for a mixture of soft and hardwood furnish.

Iowa, U.S.A. New Zealand Brazil Alberta Ontario Ontario Quebec

S1 S2 H1 H2 M1 M2 M3

a

Location

Mill codea

Table 1. List of mills sampled

Aerated lagoon – single cell Aerated lagoon – four cells Aerated lagoon – six cells Activated sludge – extended aeration Aerated lagoon Aerated lagoon Activated sludge

Effluent treatment process

230 BAKER ET AL.


231

within 48 hours. Biomass was immediately pelleted (5,000 x g for 10 minutes at 4°C) for DNA extraction and frozen at -20°C until needed. DNA Extraction and Purification Community DNA was extracted from 1.5-mL aliquots of effluent according to the method of Zhou et al. (1996). The sample was spun down and 1 mL of DNA extraction buffer (100 mM Tris-HCl, pH 8.0,100 mM EDTA, 100 mM Na phosphate [monobasic], 1.5 M NaCl, 1% cetyltrimethylammonium bromide [CTAB]) was added. The sample was vortexed to resuspend the pellet and 4 L of proteinase K (20 mg/mL) was added. The sample was mixed and rotated continuously at 37°C for 30 min. Then, 150 µL of 20% sodium dodecyl sulfate (SDS) was added and the sample was incubated at 65°C for 2 hours with shaking. The sample was then centrifuged at 6000 g for 10 min and the supernatant removed to a clean tube. To the pellet, 1 mL DNA extraction buffer and 150 µL of 20% SDS were added, vortexed for 10 sec, heated to 65°C for 10 min, centrifuged and the supernatant added to the previous supernatant. This last step was repeated once more. An equal volume (~3 mL) of chloroform/isoamyl alcohol (24:1) was added to the combined supernatants and mixed well. The sample was centrifuged (3000 x g) for 1 min and the aqueous layer (top) transferred to a clean tube. To this, 0.6 volume (~1.8 mL) of cold isopropanol was added, mixed and left on ice for 1 hour or overnight at –20°C. The sample was centrifuged (16,000 x g) for 20 min and the isopropanol poured off. The pellet was washed with 1 mL of ice-cold 70% ethanol, dried completely and dissolved in 100 to 200 µL of TE (Tris 10 mM, EDTA 0.1 mM). Humic materials were coextracted with DNA in all samples so further purification of community DNA was carried out on polyvinylpolypyrrolidone (PVPP) spin columns as described by Berthelet et al. (1996). This procedure effectively removed all coloured material from all samples. Purified DNA was dissolved in TE at a final concentration of 100 ng/mL and stored at 4°C. Amplification of DNA DNA samples were amplified in triplicate in independent reactions with each of the primer sets unless otherwise noted. 16S ribosomal RNA genes were amplified using the universal eubacterial primers fD1, 5’-AGA GTT TGA TCC TGG CTC AG-3’ and rD1, 5’-AAG GAG GTG ATC CAG CC-3’ (Weisburg et al. 1991). PCR reaction mixtures (100 µL) for 16S-RFLP analysis contained 1X PCR buffer (Roche), 250 µM of each deoxynucleotide triphosphate (dNTP) (Pharmacia), 2.5 µM of primer, 3 U of Taq polymerase (Roche), and 50 ng of template DNA. The amplification program used was: 94°C for 1 min, 50°C for 1 min, 72°C for 1 min for 36 cycles with an initial denaturation step at 94°C for 5 min and a final extension step at 72°C for 5 min. The PCR products were digested with RsaI overnight at 37°C (Moyer et al. 1996) and digestion products were separated on a 2% agarose gel in 0.5X Tris-borate-EDTA buffer at 100 volts for 6 hours.

232

BAKER ET AL.

Ribosomal intergenic spacer analysis was carried out using 1406F universal rRNA small subunit primer (5’-TGYACACACCGCCCGT-3’) and the 23SR bacterial 23S rRNA large subunit primer (5’-GGGTTBCCCCATTCRG-3’) (Borneman and Triplett 1997). PCR reaction mixtures (100 L) for 16S-RISA analysis contained 1 X PCR buffer (Roche), 250 µM of each dNTP (Pharmacia), 2.5 M of primer, 3 U of Taq polymerase (Roche), and 50 ng of template DNA. The amplification program used was: 94°C for 15 sec, 56°C for 15 sec, 72°C for 30 sec, for 30 cycles with an initial denaturation step at 94°C for 2 min and a final extension step at 72°C for 1 min. The PCR products were separated on a 2% agarose gel in 0.5X Tris-borate-EDTA buffer at 100 volts for 8 hours. Rep-PCR primers for REP sequences were, 1R, (5’IIICgICgICATCIggC-3’), and REP 2I, (5’-ICgICTTATCIggCCTAC-3’) (Versalovic et al. 1991). The PCR reaction mixture (25 L) contained 1X Gitschier buffer (83 mM [NH4]2SO4, 335 mM Tris-HCl, pH 8.8, 33.5 mM MgCl2, 33.5 µM EDTA, 150 mM -mercapto-ethanol and 800 g/mL BSA), 4 mg BSA, 10% DMSO, 1.25 mM of each dNTP (Pharmacia), 0.3 µg of each primer, 2 U of Taq polymerase (Roche) and 50 ng of template DNA (Rademaker et al. 1997, 1998). The amplification program used with the REP primers was: 94°C for 1 min, 40°C for 1 min, 65°C for 8 min for 35 cycles with an initial denaturation step at 95°C for 7 min and a final extension step at 65°C for 16 min. The Rep-PCR products were separated on a 2% Nusieve agarose gel in 0.5X Tris-borate-EDTA buffer and run at a constant voltage of 100 for approximately 16 hours. All gels were stained with ethidium bromide (0.5 µg/mL) and photographed with DS-34 fixed focal length Polaroid camera (Bio/Can Scientific). The molecular weight marker (100 base pair ladder) was purchased from Gibco/BRL and used as a reference in all the gels. Computer-Assisted Analysis of the DNA Fingerprint Patterns All three molecular methods generated complex DNA fingerprints and therefore computer assistance was essential for analysis. Polaroid pictures of gels were scanned using a BioRad (GS 700) scanner and stored as TIFF files. The images were converted, normalized against the 100 bp molecular weight marker (Gibco/BRL) and analyzed with GelCompar© software (version 4.0; Applied Maths, Kortrijk, Belgium). The “rolling disk” background subtraction method was applied. Hierarchic clustering of the whole community patterns were based on similarity and grouped using the clustering algorithm UPGMA (unweighted pair group method using arithmetic averages). Jaccard similarity coefficients were used when only DNA band positions were considered, however these may overestimate community similarities because banding intensities are ignored. Therefore we also compared patterns using Pearson correlation coefficients based on both band position and intensity (the relative areas under each of the corresponding bands). The latter have been shown to be use-


233

ful in community comparisons (Yang and Crowley 2000; Yu and Mohn 2001; Rademaker et al. 1999). It should be noted that Polaroid film and subsequent digital scanning does not accurately capture the full dynamic range of the fluorescent intensities of the bands, so similarities might be overestimated when intense bands saturate or underestimated when faint bands are simply not captured. The use of state of the art broad dynamic range CCD cameras rather than film will allow more faithful calculation of fingerprint similarities in the future. The general trends we see, however, particularly between fingerprint methods, are not likely to differ.

Results and Discussion Intra-sample (Inter-aliquot) Fingerprint Variability DNA fingerprints generated in independent PCR reactions for the same sample were 95 to 100% similar for all three methods. As shown below, independent extractions of the same samples also generated highly similar fingerprints (see Rep-PCR fingerprint comparisons below). We therefore felt confident that the degree of similarity seen between samples within and between mills reflected actual community similarity values and not differences in amplifications. All similarity values reported below were calculated by comparing the full densitogram of the fingerprint, i.e., both band and intensity were taken into account. Replicate sample similarities were always higher when only band positions were used, since some information is ignored. 16S-RFLP Fingerprint Comparisons The 16S ribosomal RNA gene is highly conserved, and different 16S-RFLP patterns are generally accepted to define different genera but may not be divergent enough to distinguish species of the same genus (Ludwig et al. 1998). It however can be useful as a phylogenetic marker to characterize microbial communities. In this study the 16S-RFLP analysis of the bacterial communities from seven pulp and paper wastewater treatment systems were performed and analyzed with Gelcompar©. A dendogram was generated for all mills. In general, samples from each mill formed individual clusters. Similarities with and between mills were derived from this dendogram, i.e., the similarity value of the node where all samples from one mill converge defines within-mill similarity, the similarity value of the node where two mills converge defines between-mill similarities. Samples from individual mills either from multiple locations within the treatment systems or from different time points were also found to generate a very high degree of similarity in the 16S-RFLP patterns (from 82–96%) suggesting that the genera within individual mills changed very little over space and time at this level of resolution. For example, in Mill H1’s treatment system that consists of 6 lagoons in tandem, samples were collected both from the edge and the middle of all

234

BAKER ET AL.

6 basins and subjected to 16S-RFLP analysis. The 16S-RFLP patterns generated from these samples differed by less than 15% from one basin to the next with the most prominent change between basins 3 and 4 (Fig. 1). This transition between basin 3 and 4 corresponds to a significant drop in both BOD and microbial numbers between basins 1 to 3 and basins 4 to 6 (mill personnel, pers. comm.). Furthermore, samples from mill S2 taken on 8 different occasions gave almost identical 16S-RFLP patterns over time with a pattern similarity of 85% (Fig. 2). This suggests that the genera were relatively stable over time within the treatment system. The comparison of the profiles between systems indicated only a 38% to 83% similarity (Table 2). Similarity between 16S-RFLP profiles from mills was partially correlated to wood furnish. The highest pattern similarity (85.3%) between mills was between M1 and M3, both of which processed a mixture of hardwood and softwood and between H1 and H2, both of which processed hardwood. The S1 and S2 mills which both processed softwood were as different from each other (43% similar) as they were to all the other mills (38–43.5% similar). Furthermore, no correlation between DNA profiles and the type of treatment process was noted. RISA Fingerprint Comparisons The RISA analysis consisted of the amplification of the intergenic spacer region between the 16S and 23S rRNA genes. The intergenic spacer region is known to contain tRNA genes in various copy numbers and exhibits length polymorphisms among bacterial genera and even within species. The variability of this region should be reflected in the complexity of the DNA banding patterns and therefore allow a finer degree of resolution than the 16S-RFLP method. Figure 3 shows several of the RISA patterns generated from 6 of the mills, some with multiple samples. Intra-mill similarity ranged from 37.3 to 92.3% while inter-mill similarity ranged from 26.2 to 65.8% (Table 2). Although RISA profiles were often very similar from different locations (Fig. 3, lanes 1–3) or from different times (lanes 10 and 11) within the same treatment system, they were sensitive enough to reflect differences (lanes 9 and 12 or lanes 14 and 15) within some of the mills. Most interestingly, mill H2, where multiple samples from the same locations (lane 4 and 6) from different time points were highly similar, showed a significantly different DNA profile from a third sample after a mill perturbation, a black liquor spill into the secondary treatment system (lane 8). This suggests that the level of resolution of the RISA analysis was sensitive enough to detect a bacterial composition shift after a shock to the system. At this level of resolution the wood furnish made little difference—there was as much dissimilarity between the mixed furnish mills as between softwood and hardwood/mixed mills and again the softwood mills (S1 and S2) were not at all similar. The type of treatment process also had no correlation to the DNA profiles obtained.

mill S1-1a

b

Fig. 1. 16S-RFLP fingerprint patterns. a) Dendogram showing the clustering of 16s-RFLP DNA fingerprints from samples from 3 mills, S1, S2 and H2; b) Samples were taken from 12 different locations within the biotreatment system of Mill H1. Lane M: Molecular weight marker; Lane 1: centre of cell 1; Lane 2: edge of cell 1; Lane 3: centre of cell 2; Lane 4: edge of cell 2; Lane 5: centre of cell 3; Lane 6: edge of cell 3; Lane 7: centre of cell 4; Lane 8: edge of cell 4; Lane 9: centre of cell 5; Lane 10: edge of cell 5; Lane 11: centre of cell 6; Lane 12: edge of cell 6.

a

mill S1-4a

mill S1-1b

mill H2-1

mill H2-6

mill H2-5

mill H2-3

mill S2-1

mill S2-4

mill S2-3


235

236

BAKER ET AL.

Fig. 2. 16S-RFLP patterns from Mill S2 taken from the same location in the biotreatment system over a 9-month period in 1997. Lane 1: April; Lane 2: May; Lane 3: August; Lane 4: September; Lane 5: October; Lane 6: November; Lane 7: December.

Rep-PCR Fingerprint Comparisons The third method, Rep-PCR, is based on primers that are complementary to naturally occurring, highly conserved, extragenic, repetitive DNA sequences throughout the genome of most bacteria. Rep sequences are found associated with 30% of bacterial operons. Amplification of DNA between Rep sites led to highly reproducible fingerprints with single isolates (Rademaker et al. 1997). Rep fingerprints of entire communities have also been found to be highly reproducible (Farhana et al. 1997; Kong et al. 2001; Matheson et al. 1997). In this study Rep fingerprints produced more complex DNA fingerprints than either 16S-RFLP or RISA patterns and allowed a greater degree of discrimination between samples. Rep-PCR amplifications from replicate samples produced highly similar fingerprints (Fig. 4a). DNA fingerprints from different samples from the same mill were more similar (75–96%) when only DNA band position was taken into account, as has been reported by other studies (Lindström 1998; Murray et al. 1998), than when both band position and intensity were considered (43–78%) (Table 2). Although similarity appears to vary considerably with this method, fingerprints among the different mill samples were still less similar than fingerprints from the same mill. When the Rep-PCR DNA patterns from the 7 different mills were compared, the percent similarity ranged from 12 to 34% (Table 2). No correlation could be seen between the fingerprints and either treatment process type or wood furnish. With Rep-PCR, samples from different locations within the same

237


Table 2. Percent average inter-sample similarities of the DNA fingerprints within and between each of the mill secondary treatment system samples using 16S-RFLP, RISA and Rep-PCR techniquesa

S1

S2

H1

H2

M1

M2

M3

S1

S2

H1

H2

M1

M2

M3

85.0 66.5 43.6

43 26.2 26.1 96.0 88.3 55.7

43.5 60.3 16.1 38 26.2 12.1 88.0 59.6 57.9

38 42.8 14.6 43.5 26.2 14.6 73.8 52.0 12.1 92.1 76.0 78.1

43.5 26.2 NAb 38 41.8 NA 61.7 26.2 NA 61.7 27.7 NA 96.0 92.3 NA

43.5 65.8 14.6 38 26.2 33.8 59.6 65.0 15.0 59.6 37.8 22.3 61.0 26.2 NA 82.2 37.3 NDc

43 28.8 12.1 38 26.2 18.6 61.7 44.6 17.5 38 28.8 15.3 85.3 26.2 NA 59.6 28.8 15.0 83.6 63.5 ND

a

The first line in each box represents the 16S-RFLP similarities, the second line represents the RISA similarities and the third line represents the Rep-PCR similarities. bNA; Not available, rep-PCR reactions were not performed with the M1 sample. cND; Not determined, only one rep-PCR reaction was performed with the M2 and M3 samples and therefore a comparison of averages from samples could not be determined.

mill or from different time points showed a lesser degree of similarity in the DNA profiles (43–78%) than with the other two methods. For example, the Rep-PCR patterns from Mill S2 from 8 different time points (Fig. 4b) showed a distinct change in the DNA pattern between the August and September samples which is in sharp contrast to the 16S-RFLP patterns generated from the same samples (Fig. 2). This difference can be attributed to the ability of the Rep-PCR method to differentiate bacteria down to the strain level, a degree of resolution not capable with the 16S-RFLP method. Whether this DNA profile change correlated to mill process para-

238

BAKER ET AL.

Fig. 3. RISA patterns of samples from 6 of the mill biotreatment communities. Lanes 1, 2 and 3: Three samples from Mill M1; Lanes 4 to 8: five samples from Mill H2, lane 4 contains a sample from the beginning of the treatment system and lane 5 a sample from the end of the system, lanes 6 and 7 are samples from a second time point 2 months later from the beginning and end of the system respectively and lane 8 represents a sample from the beginning of the system after a black liquor spill. Lane 9: a sample from Mill H1 from cell 1 of the treatment system. Lane M: 100 base pair ladder; Lane 10 and 11: two samples from Mill M3 from two time points 5 months apart; Lanes 12 and 13: two samples from Mill H1 from cells 5 and 6; Lanes 14 and 15: two samples from Mill S2 one month apart.

meters is not known since the authors are not aware of any change in mill parameters during the time that the samples were taken.

Conclusions All three methods allowed the quantification of similarity in pairwise comparisons of bacterial communities based on DNA band position and band intensity from electrophoresis of PCR products using computerassisted pattern analysis. The 16S-RFLP method however was not sensitive enough to discern bacterial community differences within a particular treatment system although different treatment systems could be differentiated. The RISA method was sensitive enough to detect some shifts in community composition within a mill treatment system. However the Rep-PCR method, which can resolve down to the strain level, showed differences in DNA patterns from one sampling time to the next that could not be resolved with either of the other two methods. This study has shown that the bacterial composition of pulp mill treatment systems can differ substantially from mill to mill—even at the


a

239

b

Fig. 4. Rep-PCR fingerprint patterns. a) Duplicate Rep-PCR fingerprints generated from independent PCR reactions from Mill S1 (lanes 1 and 2) and from Mill M2 (lanes 3 and 4); b) Rep-PCR fingerprint patterns from the Mill S2 bacterial community from 8 different time points (monthly samples over a 9-month period). Lane 1: April; Lane 2: May; Lane 3: July; Lane 4: August; Lane 5: September; Lane 6: October; Lane 7: November; Lane 8: December.

coarse level of resolution afforded by the 16S-RFLP analysis. One wonders then why some manufacturers market uniform industrial inocula products for these divergent systems and calls into questions the value of studying individual species from single mills in great detail if they are unlikely to be important in other systems. However, this study has also shown that at this level of resolution one treatment system remains fairly stable, resisting large genera changes. RISA and Rep-PCR are useful for studying the compositional changes associated with intra-mill shocks because these high resolution fingerprints have more highly variable targets relative to the ribosomal RNA genes. They can detect more subtle

240

BAKER ET AL.

species shifts. Ultimately, these tools may enable operators to be aware of community shifts that could potentially lead to sub-optimal degradation conditions and the eventual discharge of toxic effluent.

References Amann RI, Ludwig W, Schleifer KH. 1995. Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbiol. Rev. 59:143–169. Berthelet M, Whyte LG, Greer CW. 1996. Rapid, direct extraction of DNA from soils for PCR analysis using polyvinylpolypyrolidone spin columns. FEMS Microbiol. Lett. 138:17–22. Borneman J, Triplett EW. 1997. Molecular microbial diversity in soils from eastern Amazonia: evidence for unusual microorganisms and microbial population shifts associated with deforestation. Appl. Environ. Microbiol. 63:2647–2653. Bryant CW, Berkley WA. 1991. Biological dehalogenation of kraft mill wastewaters. Water Sci. Tech. 24:287–293. de Bruijn FJ, Rademaker JLW, Schneider M, Rossbach U, Louws FJ. 1996. RepPCR genomic fingerprinting of plant-associated bacteria and computerassisted phylogenetic analyses, p. 497–502. In Stacy G, Mullin B, Gresshoff PM (ed.), Proceedings of the 8th International Congress of Molecular PlantMicrobe Interactions. APS Press, St. Paul, Minn. Farhana L, DeCastro A, Allen DG, Fulthorpe RR. 1997. Community characterization methods for monitoring biofiltration. Abs. AMp17. Annual Meeting of the Canadian Society of Microbiologists, Ste Foy, Quebec. Fulthorpe RR, Liss SN, Allen DG. 1993. Characterization of bacteria isolated from a bleached Kraft mill wastewater treatment system. Can. J. Microbiol. 39:13–24. Fulthorpe RR, Rhodes AN, Tiedje JM. 1998. High levels of endemicity of 3-chlorobenzoate-degrading soil bacteria. Appl. Environ. Microbiol. 64:1620–1627. Grey JP, Herwig RP. 1996. Phylogenetic analysis of the bacterial communities in marine sediments. Appl. Environ. Microbiol. 62:4049–4059. Kong Z, Farhana L, Fulthorpe RR, Allen DG. 2001. Treatment of volatile organic compounds in a biotrickling filter under thermophilic conditions. Environ. Sci. Technol. 35:4347–4352. Kostyal E, Nurmiaho-Lassila E-L, Puhakka JA, Salkinoja-Salonen M. 1997. Nitrification, denitrification and dechlorination in bleached Kraft pulp mill wastewater. Appl. Environ. Microbiol. 47:734–741. Lindström ES. 1998. Bacterioplankton community composition in a boreal forest lake. FEMS Microb. Ecol. 27:163–174. Lindström K, Mohamed M. 1988. Selective removal of chlorinated organic compounds from Kraft mill total effluents in aerated lagoons. Nordic Pulp Paper Res. J. 1:26–33. Liss SN, Allen DG. 1992. Microbiological study of bleached Kraft pulp mill aerated lagoon. J. Pulp Paper Sci. 18:216–221. Ludwig W, Strunk O, Klugbauer N, Weizenegger M, Neumaier J, Bachleitner M, Schleifer KH. 1998. Bacterial phylogeny based on comparative sequence analysis. Electrophoresis 19:554–568.


241

Massol-Deya A, Weller R, Rios-Hernandez L, Zhou J-Z, Hickey RF, Tiedje JM. 1997. Succession and convergence of biofilm communities in fixed-film reactors treating aromatic hydrocarbons in groundwater. Appl. Environ. Microbiol. 63:270–276. Matheson VG, Munakata-Marr J, Hopkins GD, McCarty PL, Tiedje JM, Forney LJ. 1997. A novel means to develop strain-specific DNA probes for detecting bacteria in the environment. Appl. Environ. Microbiol. 63:2863–2869. Mohn WW, Wilson AE, Bicho P, Moore ER. 1999. Physiological and phylogenetic diversity of bacteria growing on resin acids. Syst. Appl. Microbiol. 22:68–78. Moyer CL, Tiedje JM, Dobbs FC, Farl DM. 1996. A computer-simulated restriction fragment length polymorphism analysis of bacterial small-subunit rRNA genes: efficacy of selected tetrameric restriction enzymes for studies of microbial diversity in nature. Appl. Environ. Microbiol. 62:2501–2507. Murray AE, Preston CM, Massana R, Taylor LT, Blakis A, Wu K, DeLong EF. 1998. Seasonal and spatial variability of bacterial and archaeal assemblages in the coastal waters near Anvers Island, Antarctica. Appl. Environ. Microbiol. 64:2585–2595. Muttray AF, Jhontang Y, Mohn WW. 2001. Population dynamics and metabolic activity of Pseudomonas abietaniphila BKME-9 within pulp mill wastewater microbial communities assayed by competitive PCR and RT-PCR. FEMS Microbiol. Ecol. 38:21–31. Rademaker JLW, de Bruijn FJ. 1997. Characterization and classification of microbes by REP-PCR genomic fingerprinting and computer assisted pattern analysis. In Caetano-Anolles G, Greshoff PM (ed.), DNA markers: protocols, applications and overviews. J. Wiley & Sons, Inc. Rademaker JLW, Louws FJ, de Bruijn FJ. 1998. Characterization of the diversity of ecologically important microbes by rep-PCR genomic fingerprinting, supplement 3, chapter 3.4.3, p. 1–26. In Akkermans DL, van Elsas JD, de Bruijn FJ (ed.), Molecular microbial ecology manual. Dordrecht, Kluwer. Rademaker JLW, Louws FJ, Rossbach U, Vinuesa P, de Bruijn FJ. 1999. Computerassisted pattern analysis of molecular fingerprints and database construction, supplement 4, chapter 7.1.3, p. 1–33. In Akkermans DL, van Elsas JD, de Bruijn FJ (ed.), Molecular microbial ecology manual. Dordrecht, Kluwer. Schneider CA, Mo K, Liss SN. 1997. Applying phenotypic fingerprinting in the management of wastewater treatment systems. Water Sci. Tech. 37:461–464. Smalla K, Wachtendorf U, Heuer H, Lui W, Forney L. 1998. Analysis of Biolog GN substrate utilization patterns by microbial communities. Appl. Environ. Microbiol. 64:1220–1225. Smit E, Leeflang P, Wernars K. 1997. Detection of shifts in microbial community structure and diversity in soils caused by copper contamination using amplified ribosomal DNA restriction analysis. FEMS Microbiol. Ecol. 23:249–261. Torsvik V, Daae FL, Sandaa R-A, Øvreas L. 1998. Novel techniques for analysing microbial diversity in natural and perturbed environments. J. Biotechnol. 64:53–62. Torsvik V, Sørheim R, Goksøyr J. 1996. Total bacterial diversity in soil and sediment communites - review. J. Ind. Microbiol. 17:170–178. Vaneechoutte M, Riegel P, De Briel D, Monteil H, Verschraegen G, De Rouck A, Claeys G. 1995. Evaluation of the applicability of amplified rDNA-restriction analysis (ARDRA) to identification of species of the genus Corynebacterium. Res. Microbiol. 146:633–641. Versalovic J, Koeuth T, Lupski JR. 1991. Distribution of repetitive DNA

242

BAKER ET AL.

sequences in and application to fingerprinting of bacterial genomes. Nucleic Acids Res. 19:6823–6831. Victorio L, Gilbride KA, Allen DG, Liss SN. 1996. Phenotypic fingerprinting of microbial communities in wastewater treatment systems. Water Res. 30:1077–1086. Wagner M, Amann R, Lemmer H, Schleifer K-H. 1993. Probing activated sludge with oligonucleotides specific for protobacteria: inadequacy of culturedependent methods for describing microbial community structure. Appl. Environ. Microbiol. 59:1520–1525. Ward DM, Bateson MM, Weller R, Ruff-Roberts AL. 1992. Ribosomal RNA analysis of microorganisms as they occur in nature. Adv. Microb. Ecology 12:219–286. Weisburg WG, Barns SM, Pelletier DA, Lane DJ. 1991. 16S ribosomal DNA amplification for phylogenetic study. J. Bacteriol. 173:697–703. Yang C-H, Crowley DE. 2000. Rhizosphere microbial community structure in relation to root location and plant iron nutritional status. Appl. Environ. Microbiol. 66:345–351. Yu Z, Mohn WW. 2001. Bacterial diversity and community structure in an aerated lagoon revealed by ribosomal intergenic spacer analyses and 16S ribosomal DNA sequencing. Appl. Environ. Microbiol. 67:1565–1574. Zhou J, Bruns MA, Tiedje JM. 1996. DNA recovery from soils of diverse composition. Appl. Environ. Microbiol. 62:316–322.


Innovative Biological Treatment Processes for Wastewater in Canada CATHERINE N. MULLIGAN1* AND BERNARD F. GIBBS2 1Department

of Building, Civil and Environmental Engineering, Concordia University, 1455 de Maisonneuve Blvd. W., Montreal, Quebec H3G 1M8 2MDS Pharma Services, 2350 Cohen St., Ville St-Laurent, Quebec H4R 2N6

Biological treatment of wastewater has been employed successfully for many types of industries. Aerobic processes have been used extensively. Production of large amounts of sludge is the main problem and methods such as biofilters and membrane bioreactors are being developed to combat this phenomenon. Anaerobic waste treatment has undergone significant developments and is now reliable with low retention times. The UASB, the original high rate anaerobic reactor, is now becoming less popular than the EGSB reactor. New developments such as the Annamox process are highly promising for nitrogen removal. For metal removal, processes such as biosorption and biosurfactants combined with ultrafiltration membranes are under development. Biosurfactants have also shown promise as dispersing agents for oil spills. If space is available, wetlands can be used to reduce biological oxygen demand (BOD), total suspended solids (TSS), nutrients and heavy metals. These innovative processes are described in this paper in terms of applications, the stage of development, and future research needs particular to Canada. Key words: biofilter, membrane bioreactor, heavy metal removal, wetlands, Annamox, biosurfactants

Introduction Biological treatment is commonly used as a secondary treatment. The major biological treatment processes for wastewater include activated sludge processes, aerated lagoons or stabilization ponds, trickling filters or fixed-film reactors, and anaerobic processes. The major groups of biological processes include aerobic, anaerobic and a combination of both. The systems are divided into suspended or attached growth processes for the removal of BOD, nitrification, denitrification, stabilization and phosphorus removal. Aerobic processes including activated sludge, trickling filters, aerated lagoons and rotating biological contactors have been used extensively. However, the supply of air is expensive in addition to the large amounts of sludge that must be sent for disposal. Recently, significant developments have been made in the area of anaerobic waste treatment. Each technology has advantages and disadvan-

* Corresponding author; [email protected]

244

MULLIGAN AND GIBBS

tages that must be considered when choosing an appropriate technology for the water treatment. The aim of this paper is to review recent developments in biological wastewater treatment, with emphasis on applications in Canada. Their applicability, state of development and research requirements will be discussed.

Aerobic Biological Treatment Aerobic Systems in Canada A recent survey was performed by CWWA for a range of facilities serving populations of various sizes (CWWA 2001). The results for the types of treatment processes are summarized in Table 1. A large number of facilities have biological systems. Activated sludge systems are the most popular followed by extended aeration. A significant number have aerobic lagoons. Overall approximately 65% of facilities responding and serving up to 5000 in population have one or more lagoons. Clearly, aerobic systems are popular in Canada for water treatment. Other treatments include primary sedimentation and chemical flocculation. This is mainly due to the significant experience of engineers in this area. This does not mean research is complete but significant improvements in aeration techniques, mixing, and sludge reduction are required. However, there are other more innovative biological wastewater treatment processes that are being developed including biofilters and membrane bioreactors. These will be discussed here.

Table 1. Survey of wastewater treatment systems in Canada (adapted from CWWA 2001)

Population Type of system Activated sludge Extended aeration Oxidation ditch RBC SBR Trickling filter Aerobic lagoons Anaerobic lagoons Facultative lagoons Total facilities

1000

>5000

13 21 1 13 1 3 31 46 59 178

59 62 10 5 5 3 117 50 52 289

53 26 2 8 6 3 44 9 30 136

>25,000 >100,000 30 3 3 2 2 6 5 1 6 55

16 1 0 2 0 2 1 0 0 30

Not given

Total

15 6 3 1 4 1 12 11 5 50

186 119 19 31 18 18 210 117 153 738

BIOLOGICAL WASTEWATER TREATMENT PROCESSES

245

Biofilters A biofilter is a new type of trickling filter that uses natural absorbents such as peat or synthetic medium for microbial attachment that can be used for the secondary and tertiary treatment of municipal or communal wastewater, restaurants, hotels, individual domestic sewage, landfill leachate (Jowett et al. 1997), food-processing wastewater and some farm animal wastewater applications. The medium (about the size of a fist) is chosen to obtain high biomass retention times, optimal microbial attachment, high porosity and separate paths for air and water. Other media such as vitrified clay called biolite was developed in France and is sold in North America in systems called the biological aerated filter. A sprinkler system distributes the liquid over the media. Loading rates are in the range of 50 to 80 cm/day which can be compared to sand filters which typically handle around 4.7 cm/day. Typical removal rates are 95% TSS, 90 to 95% BOD, 20 to 50% total nitrogen and 90 to 99% coliform bacteria. Although housings are used for cold climates, these types of reactors are mainly installed in the southern part of Canada. There are currently thousands installed in Canada, U.S.A. and Europe. Membrane Bioreactor Another variation of the activated sludge process is the incorporation of ultrafiltration membranes with a biological reactor to increase sludge retention while decreasing hydraulic retention times. In the process, wastewater enters the reactor for biological treatment. The water is then passed to the ultrafiltration step where the biomass and high molecular weight soluble components are separated from the treated water. The retained components are then recycled to the bioreactor. The process can nitrify or denitrify the wastewater when an anoxic reactor is added. Oxygen transfer rates are reduced due to fouling of the membranes. It reduces construction costs, land area, operator labour, sludge volumes, odour and chemical costs. In a typical sewage treatment plant, the sludge would be removed by the membranes, thus eliminating the need for clarifiers and sludge digestion. Organic loading rates for municipal wastewaters are typically 1 to 4 kg COD/m3-day for VSS concentrations of 15 to 25 kg/m3. However, loading rates have ranged from 0.05 to 0.66 kg BOD/m3-day with 90 to 99.8% removal efficiencies (Stephenson et al. 2000). For industrial applications, COD concentrations in the influent have been 68,000 mg/L for breweries (Kempen et al. 1997) and 29,430 mg/L for oily wastes (Zaloum et al. 1994). Sludge ages for industrial and municipal wastewaters vary between 6 to 300 days while HRTs are in the order of days for industrial wastewater and hours for municipal systems. Inert solids can accumulate in the system and thus a bleed should be incorporated into the system to remove these solids. The only drawback is that the solids are difficult to remove. Efforts are needed to

246

MULLIGAN AND GIBBS

enhance oxygen transfer. These membrane bioreactors are ideal for high salt concentrations where aggregation is difficult and for high concentrations of components that do not degrade anaerobically. More information is also required on the relationship of membranes and colloid retention. Membrane reactors can be used for many types of wastewater ranging from oily wastewater, metal finishing wastes, landfill leachates, alcohol-based cleaning solution, detergents, aqueous paint-stripping wastes, deicing fluids, soil washing effluents, contaminated groundwater to high strength and variable feed wastewater. According to Stephenson et al. (2000), 27% of membrane bioreactors have been used for industrial wastewaters, 27% for domestic wastewater, 24% for in-building, 12% for municipal and 9% for landfill leachates. There are more than 500 full-scale units installed world-wide. Most are aerobic (98%) and the rest are anaerobic. Two types of systems are utilized, the membrane external to the reactor (45%) (Fig. 1) and the membrane within the bioreactor (55%). Two other types of membrane bioreactors, membrane aeration bioreactors to enhance oxygen transfer and extractive membrane bioreactors for toxic effluents, have been developed to pilot scale only. Research and development of these units is underway. Installations of membrane bioreactors are now occurring in Canada and the United States. Three municipal installations have been commissioned since 1996 in British Columbia (capacities of 380, 1130, and 3800 m3/day) and one in Ontario (1000 m3/day) (Stephenson et al. 2000). More than 15 plants in North America have been installed for the treatment of wastewater up to 9500 m3/day. More than 15 installations have been installed for the treatment of synthetic oil and grease metal industry wastewater (up to 750 m3/day) (Coté and Thompson 2000).

Fig. 1. Schematic of a membrane bioreactor.


247

More than 50 membrane bioreactors have been installed for residential and office developments, shopping centres, schools, hotels and resorts with flows from 10 to 200 m3/day (Coté and Thompson 2000). For example, a membrane bioreactor was chosen as a resort wastewater treatment system in central Ontario (Kent 2001). An immersed hollow fibre system was used for the water generated by up to 600 guests. The system is modular and includes disinfection by UV before discharge into a lake. Sludge settling tanks are not required. Membrane surfaces are kept clean by introduction of air bubbles, flows are periodically reversed and the membranes are removed a couple of times a year and dipped in a cleaning solution. Fouling is the main concern with membrane systems and developments are still needed in this area. Other limitations include increased capital costs and maintenance over traditional activated sludge systems.

Anaerobic Processes Anaerobic processes have become popular since 1980 when the Dutch upflow anaerobic sludge blanket (UASB) system was introduced. They have several advantages over aerobic processes including lower electricity costs, high efficiencies, low construction and operating costs, low rates of sludge production, high organic loading rates production of biogas which can be used as a fuel and the aerobic microbes do not have the enzymes to remove chlorine from chlorinated compounds. Anaerobic biomass does not have to be fed continuously. Complete degradation of chlorinated compounds can be accomplished by anaerobic followed by aerobic processes. Anaerobic processes do not require oxygen and produce methane, carbon dioxide and low molecular weight end products. Methane production, once it is handled with care, can be useful for heating purposes. Sludge production is much less than for aerobic processes (5 to 20% of aerobic), reducing disposal problems and costs. All these aspects seem to potentially meet the NRC (1995) criteria for sustainable development. Degradation times may be longer, however, since anaerobic metabolism is a slower process but loading rates are higher. Anaerobic processes can transform some compounds better than aerobic ones. Chlorinated compounds in the pulp and paper industry can be dehalogenated (Parker et al. 1993) and formaldehyde can be removed (Omil et al. 1999). Macarie (1999) reviewed many recent applications that included anaerobic treatment of effluents with maleic acid, carboxymethyl-cellulose, synthetic resins and petrochemicals. Since anaerobic sludges can remain inactive for several months, seasonal wastewaters such as the fish processing or sugar refining industries, can be treated anaerobically (Omil et al. 1996). Other components that can be degraded include nitroaromatic compounds, N-substituted aromatics, alkylphenols and azo dyes (Donlon et al. 1996; Razo-Flores et al. 1996, 1997). There have been advances in high temperature treatment (van Lier 1995). Growth rates increase but granular sludge formation can be a problem. Even at low temperatures (10 to 12°C), anaerobic treatment can be

248

MULLIGAN AND GIBBS

achieved for acidified wastewater at loads up to 10 kg COD/m3-day (Rebac 1998). Acidification is required since the acidifying bacteria are affected at this temperature. An overview of the newer developments in anaerobic technologies follows. Upflow Anaerobic Sludge Blanket (UASB) Reactor A variation of the anaerobic contact process is the sludge blanket process (UASB) which is a biological tank with upflow and a settling tank developed in The Netherlands (Lettinga et al. 1980). Granules are produced during the degradation of the easily degradable organic matter and consist of high concentrations of biomass (Fig. 2). They are permanently formed and remain in the reactor. The wastewater enters the bottom of the reactor and passes through the granules. The organic matter is converted to methane and carbon dioxide and leads to the formation of gas bubbles which can provide adequate mixing and wastewater/biomass contact. The granules rise in the reactor due to the bubbles, however they will settle in the tank since their settling velocities are greater than the upflow velocity (typically 1 m/h). An adequate settling zone is provided (van Haandel and Lettinga 1994). Since the concentrations of sludge can be up to 5 to 15 kg VSS/m3, generally twice that of contact processes, recycling is not required. They are the most common type of high rate process in the world today because they can perform at higher efficiencies than anaerobic fixed film and continuous flow aerobic (Latkar and Chakrabarti 1994). Bacterial sensitivity to pH, temperature and toxic compounds, long start-up and production of odorous compounds have been cited as disadvantages for these processes. However, although chemical addition may be necessary for industrial effluent treatment, it is not usually the case for domestic wastewater and sewage (van Haandel and Lettinga 1994). The bacteria adapt well to low temperatures and can tolerate some toxicants such as aliphatic hydrocarbons and chlorinated alcohols even better than aerobic bacteria (Blum and Speece 1991). UASB reactors have been used to degrade pentachlorophenol (PCP) with up to 99% efficiency (Hendricksen et al. 1992). They have also been used for nitroaromatic compounds (Donlon 1996). Other applications include sugarbeets, fatty acids, piggery, slaughterhouse, potato starch, pulp and paper, alcohols and milk fat (McCarty 2001). Start-up times can be reduced by using adequate inoculum such as digested sludge or biomass from operating anaerobic reactors, particularly if lower operating temperatures are used (Singh et al. 1997). Toxic compounds can lead to biomass that does not settle well and subsequent biomass washout. UASB reactors are suitable for organic loads of 0.5 to 20 kg COD/m3day which is higher than aerobic processes (Kato 1994). This reduces reactor volume and space requirements. UASB reactors can be used for highstrength wastewaters with VSS:COD ratios less than 1 and with COD concentrations between 500 and 20,000 mg/L. The HRT can be less than 24 h.


a

b

c Fig. 2. Anaerobic reactors including (a) UASB, (b) Multiplate and (c) IC reactor.

249

250

MULLIGAN AND GIBBS

Cases where lower strength wastewaters with less than 1000 mg COD/L have not been frequently reported. In Kanpur, India, a full-scale UASB reactor has been treating 5000 m3 of raw sewage per day since 1989 (Draaijer et al. 1992). Another unit to treat 36,000 m3/day in the same town was subsequently built (Haskoning Consulting Engineers and Architects 1996a). Average loadings were 2.5 kg COD/m3-day with COD, BOD and TSS removals of 50 to 70%, 50 to 65% and 45 to 60%, respectively. Based on these results a pond with a one-day retention was added for installation in another Indian town (Mirzapur) (Haskoning Consulting Engineers and Architects 1996b). With loading rates of 0.95 kg COD/m3day on the UASB reactor and 0.13 kg COD/m3-day for the polishing ponds, the final effluent conditions of 30 mg COD/L and TSS of 10 mg/L could be achieved. Temperatures were between 18 and 32°C. Overall COD, BOD, TSS removal rates were 81, 86 and 89%, respectively. In Canada, a full-scale UASB system is operating at Roger’s Sugar (Taber, Alberta) for the treatment of beet sugar mill wastewater in 1999. An aerobic activated sludge reactor follows the UASB. The capacity of the system is 35,545 kg COD/day. There are numerous other applications as can be seen in Table 2. It has also been postulated that UASB reactors can be used to treat municipal wastewater in Canada, in particular for remote areas such as Indian Reserves (Singh and Viraraghavan 2000). Expanded Granular Sludge Blanket (EGSB) Reactors For treating sewage at lower temperatures (4 to 20°C), it was determined in pilot tests that the UASB reactors did not provide adequate influent distribution. The expanded granular sludge bed (EGSB) reactor was developed to incorporate higher superficial velocities (greater than 4 m/h) from the fluidized bed reactors without the need for carriers. This was accomplished by higher height to diameter ratios and the use of effluent recirculation. The higher upflow velocities expand the bed, and eliminate dead zones and lead to better wastewater/biomass contact (van der Last and Lettinga 1992). Soluble pollutants are efficiently treated by these reactors but not suspended solids or colloidal matter. A 205 m3 unit was operated in The Netherlands for approximately 33 months at temperatures between 16 and 19°C with HRT between 1.5 and 5.8 hours. COD and BOD removal rates were 30 and 40%, respectively, with no TSS removal (van der Last and Lettinga 1992). Low strength wastewaters (Kato et al. 1994) and high strength wastewaters which are diluted due to the recirculation stream are treated efficiently in these reactors. For the low strength wastewaters such as sewage, recirculation is not necessary. UASB reactors behave like static beds whereas the EGSB reactors are similar to mixed tanks (Rinzema 1988). This phenomenon increases the organic loading that can be handled by the latter type of reactor. Space requirements for the reactor are small while loading rates are high. Pumping costs, however, are increased due to the recirculation.

1994 1997

1998

Reactor of 400 m3 COD loading rate: 11.4 kg/m3 • day Reactor of 500 m3 COD loading rate: 15.4 kg/m3 • day Reactor of 50 m3 COD loading rate: 12.0 kg/m3 • day

Flow: 21 m3/h COD load: 4545 kg/day

Alcohol Flow: 42 m3/h COD load: 7680 kg/day

Chemical Flow: 0.4 m3/h COD load: 300 kg/day

Canning operation

Commercial alcohol

3M

1994

Reactor of 800 m3 COD loading rate: 13.6 kg/m3 • day

Flow: 42 m3/h COD load: 10,900 kg/day

A. Lassonde. Inc.

1993

Reactor of 500 m3 COD loading rate: 14.3 kg/m3 • day

Year of start-up

Volume of reactor and loading

Yeast Flow: 17 m3/h COD load: 7173 kg/day

Type of wastewater

Fleishmann’s yeast

Installation

Table 2. Anaerobic UASB plants installed in Canada in the last 10 years (Biothane 2002) BIOLOGICAL WASTEWATER TREATMENT PROCESSES

251

252

MULLIGAN AND GIBBS

The internal circulation (IC) reactor technology is a type of EGSB reactor. It consists of two UASB reactors, one on top of the other (Fig. 2). One is for high loads while the second is for low loads. The gas from the first stage drives a gas lift and internal circulation. The biogas is collected in the top of the reactor. The four basic processes in the reactor are the mixing section, expanded bed section, the polishing section and the recirculation system. The Biobed system, another type of EGSB reactor, has been completed at Robin Hood Multi-Foods near St. Catherine and is under construction at McCain Foods in Grand Falls, N.B., and will be used as a pretreatment at Inter-Quisa in Montreal. More than 40 systems are operating or are under construction throughout the world. Therefore, activity in Canada has been strong. Applications are for the brewery, chemical, fermentation and pharmaceutical industries. The Multiplate Reactor The multiplate reactor technology was constructed in 1991 at a dairy plant in the province of Quebec (Canada) (Mulligan et al. 1996). The reactor is comprised of a shell, plates, parallel feed entrances and lateral gas exits (Fig. 2). Typical total COD removal rates of greater than 93% and soluble COD removal rates of 98% are achieved. Other effluents have also been pilot tested with this reactor for volatile organic compounds (VOCs), aircraft deicing agents, brewery and potato processing wastewater (Mulligan et al. 1997). Its potential at the laboratory scale for the degradation of chlorinated solvents is being investigated. More full-scale demonstrations of this unit are required. Anaerobic Sequencing Batch Reactor The Anaerobic Sequencing Batch Reactor (ASBR) is an anaerobic version of the conventional SBR technology. It is applicable for high strength wastewaters and can remove 75 to 94% COD with hydraulic retention times of 8 to 24 hours. The age of the biomass is 60 to 70 days. The four cycles of fill, react, settle and decant operate on three- to twelve-hour cycles. Operation is based on timing. Due to the batch-fed operation, short-circuiting does not occur. The biomass is highly granulated and contains many bacterial species and fungi with mineral deposits. These granules settle rapidly at a rate of a metre per minute. Organic loading rates of 4 kg COD/m3-day are used (Beun et al. 1999). Dilution of toxic materials does not occur. This type of reactor appears to still be under development due to a lack of full-scale systems. A semi-commercial system has been developed by Agriculture and Agri-Food Canada for swine manure slurries and has been pilot tested at 30°C for the treatment of slaughterhouse wastewater (Massé and Masse 2000). Annamox Process Recently, an innovative process, known as Anammox (anaerobic ammonium oxidation), has been discovered (Strous et al. 1999). Up to


253

2.6 kg total N/m3 reactor-day can be achieved compared to 0.1 kg total N/m3 reactor-day for activated sludge processes (Jetten et al. 1999; STOWA 1996). SBR and fluidized bed reactors can be used. This process can convert ammonium in the wastewater to nitrogen gas under anoxic conditions with nitrite as the electron acceptor and ammonium as the electron donor with sludge production. Sludge generation in this process is very low. This reaction is very promising but insufficient work has been done to take advantage of this process. The main disadvantage is the slow doubling time of Annamox bacteria (11 days). The ratio of ammonium to nitrite should be 1:1.3 and has been accomplished by partially treating the wastewater with the Sharon reactor (1 day HRT at 35°C). In our research (Mulligan and Chan 2001), the feasibility of ammonium removal from wastewater at the same time as COD removal by nitrite addition has been examined at room temperature. Further experiments will also be needed to determine the ratio of nitrite to ammonia required for the Anammox process to take place. Using continuous reactors may favour their development. Development of this technology will be highly important for the anaerobic treatment of municipal and other industrial wastewaters where nitrogen removal is essential. Analysis of Anaerobic Technologies Full-scale anaerobic systems have been quite successful. There are many misconceptions about anaerobic treatment that limit its use. These problems have arisen due to poor designs before the 1950s and a lack of understanding of anaerobic processes. A major advantage for anaerobic processes over aerobic is the decreased rate of sludge production. Sludge production can be between three and twenty times less than for aerobic processes (Rittmann and Baskin 1985). The costs of disposal of large amounts of sludge can be substantial. In addition, the processing of the sludge before disposal is energy intensive unless gravity or flotation thickening is feasible. Overall, proper reactor design and operation can overcome any disadvantages of anaerobic treatment. Since anaerobic treatment processes often lead to effluents that are greater than regulatory requirements, further treatment is often necessary. Aerobic polishing can be used. This maintains the advantages of anaerobic treatment with an effluent that meets requirements for discharge into streams or rivers. The solids from the aerobic treatment can be treated in the anaerobic system. An analysis of full-scale anaerobic technologies was performed by Frankin (2001). He showed that there were approximately 1215 plants in 65 countries. A breakdown of the processes is shown in Table 3. Although UASB systems are still the most common, the growth of the EGSB in the last few years is particularly noteworthy. The average design loading rates are also shown which clearly indicate that the EGSB design is twice that of the UASB, thus leading to smaller reactors. This would also indicate why the UASB reactors are gradually being replaced by the EGSB. In terms of applications, food wastewaters are the most treated by anaerobic

254

MULLIGAN AND GIBBS

Table 3. Distribution of full-scale anaerobic technologies and their average organic loads (Frankin 2001)

Process Low rate UASB Fixed bed Fluidized bed Hybrid EGSB Unknown

% of Total plants in database

% of Total plants built from 1990 to 1996

% of Plants built from 1997 to 2000

Average load (kg COD m3 • day)

15 56 4 1 1 16 5

12 68 4 2 1 8 6

8 34 3 1 2 50 3

2 10 unknown unknown 7.5 20 unknown

reactors, followed by breweries and beverages and then distilleries and fermentation (Fig. 3). The newest development of anaerobic systems is for municipal wastewater. Until now, applications at low temperatures have been limited due to low treatment rates. The hydrolysis step will need to be improved, however. Costs are more competitive than activated sludge treatment when sewage temperatures are above 15°C and the land cost is above $23/m2 (Hulshoff Pol et al. 1998). The same trend as for other countries has held in Canada where the pulp and paper and the food industries are the most common applications for anaerobic treatment. Japan, Germany, The Netherlands and the United States are leading countries in implementing anaerobic reactors for wastewater treatment. Canada is tenth as a country in applying anaerobic technologies for wastewater treatment at 0.8 reactors per million habitants (Hulshoff Pol et al. 1998). There are no plants that use anaerobic treatment for municipal wastewater treatment in Canada, despite the potential benefits in this area. The most appropriate anaerobic technology should be selected based on bench or pilot tests with the actual wastewater, particularly if the wastewater contains toxic components. The result of the tests depends on retaining an active microbial population. Process conditions should be designed and operated for the optimal performance of the microorganisms. More development is needed in modelling and in the control of anaerobic reactors and the use of biosensors in the presence of high solids contents (Aubrun et al. 2001). Choice of the inoculum sludge is very important, in addition to the training of the operators. The main factors for consideration are easy construction, reliability over a variety of wastewater characteristics, easy restart, low operating costs and high efficiency. UASB and contact processes are generally more simple to operate and maintain while EGSB, IC and fluidized bed reactors are more complex. Developments in the treatment of recalcitrant compounds


255

and the use of low temperature processes will enable wider adoption of anaerobic technologies.

Wetland Systems Natural wetlands are areas of land with the water surface near that of the land, thus maintaining saturated soil conditions and vegetation including plants, peat, wildlife, microbial cultures and vegetation including cattails (Typha spp.), reeds (Phragmites spp.), sedges (Carex spp.), bulrushes (Scirpus spp.), rushes (Juncus spp.), water hyacinths (Eichhornia crassipes), duckweeds (Lemna spp.), grasses and others detailed by Mitsch and Gosselink (1992). Constructed wetlands have been specifically designed to include these species for the removal of BOD, SS, nutrients and heavy metals for optimal performance. Denitrification also occurs due to the anaerobic conditions in the water. It was reported by Reed et al. (1995), that 1000 managed wetlands are in operation throughout the world. The wetland systems can be designed either as surface flow with a free water surface or as a subsurface flow (Fig. 4) where the water must enter after passing through a permeable medium. Both types of wetland systems can be applied to commercial and industrial wastewaters. High strength wastewaters are usually treated anaerobically first. Both types of wetlands have been used for wastewater from food processing, pulp and paper processing, chemical production and oil refineries. Pilot tests may be preferable if inadequate treatment data exists. Wetlands have also been used for the removal of sediments from urban stormwater from landscapes, streets and parking lots. They also have some benefit for BOD, nitrate, phosphate and trace metal removal. The basis of these systems is a combination of shallow marshes and deep ponds to which wet meadows and shrub areas can be added. In a partial list of the wetlands in the North American Wetland Treatment System Database, 154 were for the treatment of municipal wastewater, nine for industrial, six for agricultural wastewater and seven

Breweries and beverages Distilleries and fermentation Chemical Pulp and paper Food Landfill leachate Unknown

Fig. 3. Breakdown of applications for anaerobic treatment.

256

MULLIGAN AND GIBBS

Fig. 4. Schematic of wetland systems including surface flow and subsurface flow.

were for stormwater (Kadlec and Knight 1996). These systems treat more than 190 m3/day. Of these, 120 are surface flow systems and 48 are subsurface systems, while 8 are both. Wetland treatment systems are, thus, becoming more and more accepted for municipal, agricultural and industrial wastewaters, as well as for stormwater management. Wetlands are becoming more accepted for stormwater quality management in Canada (Warner and Li 2000). Two wetlands were constructed in Nova Scotia for domestic wastewater treatment and as a wildlife habitat. Rates of removal for BOD, suspended solids, fecal coliform bacteria, total phosphorus and total nitrogen were 90% from 1996 to 1998 (Hanson and McCullough 2002). In general, average removal efficiencies are 50 to 80% as shown in Table 4 (API 1998). Phenols can be reduced by 70% in the petroleum industry and VOCs up to 95%. Even 50% metal removal efficiencies have been achieved for aluminum, cadmium, copper, iron, lead, mercury, nickel, silver and zinc. This is due to growing confidence in the performance of these systems and the shortage of affordable technologies (Kadlec and Knight 1996). For treatment of acid mine drainage, constructed wetlands could be used in Canada (MEND 1999). Design will need to be optimized and

257


Table 4. Summary of the performance of North American wetland treatment systemsa

Concentration (mg/L) Parameter BOD TSS NH4-N Total N Total P aAdapted

In

Out

Efficiency (%)

29.8 46.0 4.97 9.67 3.8

8.1 13.0 2.41 4.53 1.68

73 72 52 53 56

from Kadlec and Knight (1996) and API (1998).

many parameters will need to be understood including biological reaction and metal retention in the wetlands. The effect of freezing temperatures and variable flows will need to be addressed. The wetlands will most likely be applied where winters are short and mild and where flow rates are constant. If operation during the winter is not possible, large retention ponds will be required during this period. Capital costs have been estimated at $8 to 48/m2 of wetland with $85,000 in operating costs for a 60 L/min wetland (MEND 1999). In the Biosphere in Montreal, Quebec, wastewater has been treated since 1995 with a pilot wetland system part of the year (Environment Canada 2002). It consists of three ponds with an area of 800 m2. Total retention time is 2 weeks. The system is working well and is an example for future housing projects. More intensive monitoring of the system will be taking place. A more recent approach which could be applied in Canada was developed in France for communities with less than 1000 inhabitants (Betts 2002). It uses chrysanthemum plants for aerobic wetland treatment of the wastewater. Ammonia can be easily treated within 72 hours. Nitrogen and phosphorus removal was 40 to 80%, BOD removal was 91% and 95% of the suspended solids were removed. Heavy metal removal is unknown at this point and more data is required to determine the effects of cold weather that may freeze the plants. For the use of wetlands to develop, more collaboration will be required between engineers and scientists due to the interdisciplinary nature of the technology. More than 100 natural wetlands exist in Canada (Warner and Li 2000). Constructed wetlands will enable engineers to design systems with more control and obtain regulatory approval. Capital costs are low, with low operation and maintenance requirements. This technology is emerging and its advantages make it attractive. It will need everyone (engineers and scientists) to share the knowledge to improve

258

MULLIGAN AND GIBBS

know-how. An approach similar to the U.S. may be necessary where interdisciplinary groups help to develop guidelines (Cole 1998). Currently, no accepted methodology exists for wetland design. Depths of 1.5 m and length to width ratios of 3:1 with surface loadings of up to 220 kg BOD/ha-day usually provide good BOD and SS removal results (Rittmann and McCarty 2001). This is a sustainable and green technology for purifying water with numerous benefits for the future.

Metal Treatment Processes Biosorption Biosorption involves the removal of metals from wastewater via adsorption on living or dead biomass. The biomass can include bacteria (Bacillus subtilis, Bacillus licheniformis), yeast (Candida tropicalis), fungus (Aspergillus niger, Penicillium chryosogenum, Rhizopus arrhizus), algae (Sargassum natans, Ascophyllum rodosum, Fucus vesiculosus) and plant material (peat moss, wood chips and pine cones). Algal biomass (Sargassum natans) can uptake metals up to almost 40% of its dry weight. Many metals can be adsorbed such as silver, gold, cadmium, chromium, copper, iron, mercury, nickel, lead, zinc and radioactive metals. Various types of materials have been used for immobilization including alginate, polyacrylamide, polysulfone, silica gel, cellulose and glutaraldehyde. For commercial systems, biosorption processes can take place in batch or continuous-stirred tanks, fixed packed beds, and fluidized beds (Volesky 1990). Biosorbent use depends on biosorption capacity, availability of the biosorbent, cost, ease of regeneration and use in various reactor configurations. Eluents such as dilute acids or carbonate can be used to desorb the adsorbed contaminants. Aspergillus, Penicillium and Saccharomyces can withstand 10 cycles of regeneration without decreased adsorption capacity. However, more research is needed in this area. Other biosorbents such as anaerobic sludge is currently being evaluated for the removal of Cd, Cu, Pb and Ni (in batch and continuous systems (Alhawari and Mulligan 2002). Biosorption can be useful for radionuclides from dilute streams such as mine leachates. Aspergillus niger can adsorb between 31 to 214 mg/g of uranium, Rhizopus arrhizus can adsorb about 200 mg/g and Saccharomyces cerevisiae can adsorb 150 mg/g. This can be compared to traditional adsorbers such as ion exchange resin IRA-400 (79 mg/g) and Activated Carbon F-400 (145 mg/g). Only a few biosorbents have been commercialized. Potential applications of these biosorbents include industrial effluent polishing and metal removal from dilute effluents. The advantages of using biosorbents include versatility and flexibility, robustness, selectivity of heavy metals over alkaline earth metals, ability to reduce metal concentrations to drinking water standards, and cost-effectiveness (Garnham et al. 1992). Current constraints are the competition with ion exchange


259

resins, the low capacity of metal fixation in terms of mg of metal adsorbed per gram of sorbent, the selectivity and the ability to regenerate these materials. Engineers, in general, prefer more established processes such as ion exchange as they have a lack of knowledge concerning biosorbents. A greater understanding of these biological-based systems may help their commercialization potential in the future. There have been a few studies related to acid mine drainage and biosorption. The use of living biomass is not likely to be feasible in the winter. Dead biomass systems appear more promising. Biosorption could also be used as a secondary treatment process (MEND 1999). Metal Precipitation and Sulfur Removal Sulfate-reducing bacteria include Desulfovibrio, Desulfotamachulum, Desulfobacter, Desulfococcus, Desulfonema and Desulfosarcina. They can remove metals by hydrogen sulfide production which subsequently precipitates metals. Sulfur-oxidizing bacteria include Thiobacillus thioxidans, T. thioplaus and T. denitrificans. Reactors have been developed to take advantage of these processes. For example, for heavy metal and sulfate removal, metal precipitates are formed. In a second reactor, the sulfide is converted to sulfur. Metal reuse is possible if only one metal is used such as zinc. These biological processes have been operated in the treatment of electronic component and electro-plating wastewaters. In the past, sulfur effluents have been treated with lime which forms gypsum that has to be landfilled. Although lime addition is simple, the sludge is not easy to dewater and frequently sulfate concentration below 1500 mg/L cannot be achieved. The final sulfur product contains 60% solids with a purity of 95% and can be used for sulfuric acid production or as soil amendments. An example of a full-scale installation is at a synthetic fiber production plant (Emmen, The Netherlands) where 40 m3/h of wastewater containing 2 g/L sulfate has been treated since 1995. Approximately 75% of the sulfate is converted to sulfur. Currently through oxidative or reductive processes, full-scale reactors up to 2000 m3 have been constructed. More than 24 commercial plants have been constructed for desulphurization and six combine metal and sulfur removal. In the future, developments will be required for removal of mercaptan and other organic sulfur compounds (Kuenen and Lens 2001). Reactor systems for sulfate-reducing bacteria have been pilot tested for treatment of acid mine drainage in Canada and appear feasible for low flow rates (MEND 1999). Longer term studies at flow rates higher than 1 L/min will be required. The choice of carbon source will be the key to the success of the reactor. Open reactors can only be used where the winter is not exceedingly cold. Closed reactors, however, could be used in all climates. It has been estimated that small open systems of 50 to 60 L/min could cost approximately $34,000 and closed systems (75 to 100 L/min) approximately $56,000 (MEND 1999). Recently, the BioSulphide/Thiopaq

260

MULLIGAN AND GIBBS

process (Fig. 5), the result of a Canadian-Netherlands agreement enabled a commercial plant to be designed, constructed and commissioned at the Caribou Mine in Bathurst, New Brunswick, for the treatment of acid water and selected metals including copper, zinc, cadmium and lead (BioteQ 2002).

Use of Biosurfactants Biosurfactants, surface active agents produced by bacteria or yeast, are potentially useful in wastewater treatment, particularly due to their anionic nature, low toxicity, biodegradability and excellent surface active properties. Recently, their feasibility for enhancing metal removal has been demonstrated (Mulligan et al. 2001). Copper and zinc (10 mg/L) were rejected by ultrafiltration membranes and with membranes of molecular weight cutoffs of 50,000 amu for surfactin and 10,000 amu for rhamnolipid. Concentrations of greater than 0.1% for both surfactants showed the highest metal rejection ratios (greater than 80%). This phenomenon is due to metal complexation with the biosurfactants. Further experiments are now being performed with hollow fiber membranes. Another recent development is the feasibility of biosurfactants for dispersing oil slicks (Holakoo and Mulligan 2001). At 25°C and a salinity of 35%°, a solution of 2% rhamnolipids diluted in saline water and applied at a dispersant to oil ratio (DOR) of 1:2, could immediately disperse 65% of a Brut crude oil. Co-addition of 60% ethanol and 32% octanol with 8% rhamnolipids applied at a DOR of 1:8 improved dispersion to 82%. Dispersion efficiency decreased in fresh water and at lower temperatures but altering the formulation could improve efficiencies. Comparison of the dispersion behaviour to Corexit showed that the rhamnolipids had excellent potential as non-toxic oil dispersing agents.

Future Developments Many types of wastewater can be treated biologically with proper analysis and environmental control. Changes in the environment must allow the organisms to adapt or the effects may be highly detrimental. In the future, research must focus on the development of systems that can increase the rate of the treatment process to decrease retention times and subsequently reactor volumes. There is usually resistance by engineers, and waste treatment plant operators, among others, to biological augmentation (addition of specific microorganisms). Experts in design, operation and biological processes will need to combine their efforts to enhance system performance, particularly for the treatment of recalcitrant compounds. Because experience is fairly limited in the biological treatment of toxic compounds, it is difficult to predict their fate and effect in bioreactors. More developments are required for thermophilic anaerobic

261

Fig. 5. Reactor for removal of sulfur.


262

MULLIGAN AND GIBBS

reactors, granular SBR and membrane reactors, in addition to odour control for anaerobic reactors, and improving efficiencies of refractory organic degradation. Systems such as wetlands are highly complex and research is needed to determine specific mechanisms for toxicity reduction. Due to the success of sulfate and metal removal in the Netherlands, installations will be expected in Canada. Engineers will need to have a better understanding of biological processes through working with multidisciplinary teams. For the future, anaerobic treatment and other biological treatment processes such as constructed wetlands should play an important role in the sustainable development of water resources. Fresh water is depleting rapidly in countries such as India, China and even the U.S. and Canada. Rivers are drying up and water table levels are decreasing. Therefore, we must protect the quality of the water that we have by treating the water in the most appropriate manner before discharge. These processes will need to be efficient and cost-effective and must not generate further waste problems. In summary, the increasing population is leading to fewer waste management options, environmental destruction, and increased disasters due to global warming. Environmental management and technological development of biological processes should clearly be a priority.

References Alhawari A, Mulligan C. 2002. Biosorption of lead, cadmium, copper and nickel by anaerobic sludge. Presented at the 18th Eastern Canadian Symposium on Water Quality Research (CAWQ), October 18, Montreal, Canada. American Petroleum Industry (API). 1998. Treatment wetlands for the petroleum industry, Brochure prepared for the API Biomonitoring Task Force by CH2M Hill, Washington, D.C. Aubrun C, Theilliol D, Harmand J, Steyer JP. 2001. Software sensor design for COD estimation in an anaerobic fluidised bed reactor. Water Sci. Technol. 43:115–122. Betts KS. 2002. Beautifying wastewater treatment. Env. Sci. Technol. 36(11): 251A–252A. Beun JJ, Hendricks A, van Loosdrecht MCM, Morgentoth E, Wilderer P, Heijnen JJ. 1999. Anaerobic granulation in sequencing batch reactor. Water Res. 33:2283–2290. BioteQ. 2002. The Biosulphide/Thiopaq process. The BioteQ Environmental Technologies Inc., Vancouver, B.C., http://www.bioteq.ca. Biothane. 2002. UASB reactors in Canada. http://www.biothane.com/ map_files/canada.html. Blum DJW, Speece RE. 1991. A database of chemical toxicity to environmental bacteria and its use in interspecies comparisons and correlations. Res. J. WPCF 63(3):198–207. Canadian Water and Wastewater Association (CWWA). 2001. National survey of wastewater treatment plants. Final Report, June 14. Cole S. 1998. The emergence of treatment wetlands. Environ. Sci. Technol. 32:218A–223A. Coté P, Thompson D. 2000. Wastewater treatment using membranes: the North American experience. Water Sci. Technol. 41:209–215.


263

Donlon BA, Razo-Flores E, Lettinga G, Field JA. 1997. Continuous detoxification, transformation and degradation of nitrophenols in upflow anaerobic sludge blanket reactors. Biotech. Bioeng. 51:439–449. Draaijer H, Maas JAW, Schaaoman JE, Khan A. 1992. Performance of the 5 MLD UASB reactor for sewage treatment at Kanpur, India. Water Sci. Technol. 25(7):123–133. Environment Canada. 2002. The Biosphere’s natural wastewater treatment plant. http://biosphere.ec.gc.ca/cea/actu/doss/doss_00024_a.html. Fox P, Suidan MT, Bandy JT. 1990. A comparison of media types in acetate fed expanded-bed anaerobic reactors. Water Res. 24(7):827–835. Frankin RJ. 2001. Full-scale experiences with anaerobic treatment of industrial wastewater. Water Sci. Technol. 44:1–6. Garnham GW, Codd GA, Gadd GM. 1992. Accumulation of cobalt, zinc, and manganese by the estuarine green microalga Chlorella salina immobilized in alginate microbeads. Environ. Sci. Technol. 26:1764–1770. Hanson A, McCullough B. 2002. Constructed wetlands for wastewater treatment and wildlife habitat: seasonal variation. http://auracom.com/~bacap/ Presentations/A.Hanson.htm. Haskoning Consulting Engineers and Architects. 1996a. 36 MLD UASB treatment plant in Kanpur, India, Evaluation Report on Process Performance. Internal Report. Haskoning Consulting Engineers and Architects. 1996b. 14 MLD UASB treatment plant in Mirzapur, India, Evaluation Report on Process Performance. Internal Report. Hendrikson HV, Larsen S, Ahrung BK. 1992. Anaerobic dechlorination of pentachlorophenol in fixed film and upflow anaerobic sludge reactors using different inocula. Biodegradation 3:399–405. Holakoo L, Mulligan CN. 2002. On the capability of rhamnolipids for oil spill control of surface water. Proceedings of the Annual Conference of the Canadian Society for Civil Engineering, June 5–8, Montreal, Canada. Hulshoff Pol L, Euler H, Schroth S, Wittur T, Grohganz D. 1998. GTZ sectoral project “promotion of anaerobic technology for the treatment of municipal and industrial wastes and wastewater.” Proc. 5th Latin-American WorkshopSeminar “Wastewater anaerobic treatment,” 27–30 October, Vina del Mar, Chile. Jetten MSM, Strous M, van de Pas-Schoonen KT, Schalk J, Van Dougen LGJM, Van de Graaf AA, Logeman S, Muyzer G, Van Loosdrecht MCM, Kuenen JG. 1999. The anaerobic oxidation of ammonium. FEMS Microbiol. Rev. 22:421–437. Jowett EC, Cook WG, Prentice GL, Hughes CP, Ford FC, Klint S. 1997. Removal of VOCs from landfill leachate using absorbent aerobic biofiltration. American Chemical Society, Ninth Annual Symposium on Emerging Technologies in Hazardous Waste Management, Pittsburgh, September 15–17, 1997. Kadlec RH, Knight RL. 1996. Treatment wetlands. CRC Lewis Publishers, Boca Raton. Kato MT, Field JA, Kleerebezem R, Lettinga G. 1994. Treatment of low strength soluble wastewater in UASB reactors. J. Ferm. Bioeng. 77(6):679–686. Kempen R, Draaijer H, Postma H. 1997. A membrane bioreactor for industrial effluent MBRI-Proc. 1st Intl. Mtg. on Membrane Bioreactors for Wastewater Treatment, Cranfield University, Cranfield, UK. 8 p. Kent C. 2001. Ontario resort upgrades its wastewater treatment facility. Env. Sci. Eng. Jan. http://www.esemag.com/0101/resort.html.

264

MULLIGAN AND GIBBS

Kuenen G, Lens PNL. 2001. Biological S-cycle – the biological sulfur cycle: novel opportunities for environmental biotechnology. Water Sci. Technol. 44:57–65. Latkar M, Chakrabarti T. 1994. Performance of upflow anaerobic sludge blanket reactor carrying out biological hydrolysis of urea. Water Environ. Res. 66(1):12–15. Lettinga GA, van Velsen AF, Hobma SW, de Zeeuw W, Klapwijk A. 1980. Use of the upflow sludge blanket (USB) reactor concept for biological wastewater treatment, especially for anaerobic treatment. Biotech. Bioeng. 22(4):699–734. Macarie H. 1999. Overview of the application of anaerobic digestion to the treatment of chemical and petrochemical wastewaters, p. 405–412. Proceedings of IAWQ Symposium on Waste Minimization and End of Pipe Treatment in Chemical and Petrochemicals Industries. 14–18 November, Merida, Yucatan, Mexico. Massé DI, Masse L. 2000. Treatment of slaughterhouse wastewater in anaerobic sequencing batch reactor. Can. Agr. Eng. 42:131–137. McCarty PL. 2001. The development of anaerobic treatment and its future. Water Sci. Technol. 44:149–156. MEND. 1999. Review of passive systems for treatment of acid mine drainage, MEND Report 3.14.1, CANMET, NRCan, May 1996, Revised 1999. Metcalf and Eddy Inc. 1991. Wastewater engineering: treatment, disposal and reuse, Third Edition. Tchobanaglous G, Burton F (ed.), McGraw–Hill, New York. Mitsch WJ, Gosselink JG. 1992. Wetlands, 2nd Ed. Van Nostrand Reinhold, New York. Mulligan CN, Chan TY. 2001. Developments in anaerobic wastewater treatment. Proceedings of the 29th Annual Conference for the Canadian Society for Civil Engineering May 30–June 2, 2001, Victoria, B.C. Mulligan C, Chebib J, Safi B. 1997. Anaerobic treatment using the multiplate reactor. Can. Environ. Prot. 9(7):16–17. Mulligan CN, Safi B, Mercier P, Chebib J. 1996. Full scale treatment of dairy wastewater using the SNC multiplate reactor, p. 544–556. In Moo-Young M, Anderson WA, Chakrabarty AM (ed.), Environmental biotechnology. Kluwer Academic Publishers, Dordrecht. Mulligan CN, Yong RN, Gibbs BF. 2001. Heavy metal removal from sediments by biosurfactants. J. Haz. Mat. 85:111–125. NRC. 1995. The role of technology in environmentally sustainable development. National Academy Press. Washington, D.C. Omil F, Mendez D, Vidal G, Mendez R, Lema JM. 1999. Biodegradation of formaldehyde under anaerobic conditions. Enzyme Microbiol. Technol. 24:255–262. Omil F, Mendez R, Lema JM. 1996. Anaerobic treatment of seafood processing wastewaters in an industrial pilot plant. Water S.A. 22(2):173–181. Parker WJ, Hall ER, Farquar GJ. 1993. Assessment of design and operating parameters for high rate anaerobic dechlorination of segregated Kraft mill bleach plant effluents. Water Environ. Res. 65(3):264–270. Razo-Flores E, Luijten M, Donlon BA, Lettinga GA, Field JA. 1996. Biotransformation and biodegradation of N-substituted aromatics in methanogenic granular sludge. FEMS Microbiol. Rev. 20(3):526–538. Razo-Flores E, Luijten M, Donlon BA, Lettinga GA, Field JA. 1997. Biodegradation of selected azo-dyes under methanogenic conditions. Water Sci. Technol. 36(6):65–75.


265

Rebac S. 1998. Psychrophilic anaerobic treatment of low strength wastewaters. Ph.D. Thesis, Wageninen, Agricultural University, Wageningen, The Netherlands. Reed SC, Crites RC, Middlebrooks EJ. 1995. Natural systems for waste management and treatment. McGraw-Hill, New York. Rinzema A. 1988. Anaerobic treatment of wastewater with high concentrations of lipids or sulfate. Ph.D. Thesis, Wageningen Agricultural University, Wageningen, The Netherlands. Rittmann BE, Baskin DE. 1985. Theoretical and modelling aspects of anaerobic treatment of sewage, p. 55–94. In Switzenbaum MS (ed.), Proceedings of the Seminar/Workshop on Anaerobic Treatment of Sewage, Amherst, U.S.A. Ritmann BE, McCarty PL. 2002. Environmental biotechnology: principles and applications. McGraw Hill, Boston. Singh KS, Viraraghavan T. 2000. Performance of UASB at 6 to 32oC for municipal wastewater treatment. Water Qual. Res. J. Canada 35:113–124. Singh KS, Viraraghavan T, Karthikeyan S, Caldwell DE. 1997. Low temperature start-up of UASB reactors for municipal wastewater treatment. Proc. 8th Int. Conf. Anaerobic Digestion. Sendai, Japan 3:192–195. Stephenson T, Judd S, Jefferson B, Brindle K. 2000. Membrane bioreactors for wastewater treatment. IWA Publishing, London. STOWA. 1996. Removal of ammonium from sludge water with the Annamox process. Feasibility study. Report no. 96–21. STOWA, Utrecht, The Netherlands, ISBN 9– 744476 554. Strous M, Kuenen G, Jetten MSM. 1999. Key physiology of anaerobic ammonium oxidation. Appl. Environ. Microbiol. 65:3248–3250. van der Last ARM, Lettinga G. 1992. Anaerobic treatment of domestic sewage under moderate climatic (Dutch) conditions using upflow reactors at increased superficial velocities. Water Sci. Technol. 25(7):167–178. Van Haandel AC, Lettinga G. 1994. Anaerobic sewage treatment. A Practical Guide for Regions with a Hot Climate. John Wiley and Sons Ltd., Chichester. van Lier JB. 1995. Thermophilic anaerobic wastewater treatment: temperature aspects and process stability. Ph.D. Thesis, Wageninen, Agricultural University, Wageningen, The Netherlands. Volesky B. 1990. Removal and recovery of heavy metals by biosorption, p. 8–43. In Volesky B (ed.), Biosorption of heavy metals. CRC Press, Boca Raton. Warner BG, Li J. 2000. Application of wetland technologies for municipal stormwater management, p. 510–514. In 6th Environmental Engineering Specialty Conference of the CSCE & 2nd Spring Conference of the Geoenvironment Division of the Canadian Geotechnical Society. June 7 to 10, London, Ont. Zaloum R, Lessard S, Mourato D, Carriere J. 1994. Membrane bioreactor treatment of oily wastes from a metal transformation mill. Water Sci. Technol. 30(9):21–27.


Binding of Hydrophobic Organic Contaminants to Humalite-Derived Aqueous Humic Products, with Implications for Remediation DALE R. VAN STEMPVOORT,* SUZANNE LESAGE AND HELENA STEER

National Water Research Institute, P.O. Box 5050, Burlington, Ontario, Canada L7R 4A6

In order to assess the potential of commercial humic products in environmental remediation, their binding to heptachlor, hexachlorobenzene, hexachloroethane, 1,2,4-trichlorobenzene and n-hexane was measured. The binding was similar in tests with two colloidal-phase humic products prepared from humalite (naturally occurring, oxidized organic deposits found adjacent to coal) from Alberta, Canada (Luscar Ltd.). There were minor or negligible changes in binding strength for these two products over the full range of aqueous concentrations tested (1.67 to 33.4 g organic C/L), and for Aldrich humic acid. Aldrich humic acid was a stronger binder of the hydrophobic organic compounds than the two Luscar humic products, by factors of approximately 3 to 7. The binding of hydrophobic organics to the two Luscar humic products was similar in strength to that predicted for humic substances that occur in natural aquatic environments. Our binding data suggest that the use of concentrated Luscar humic products as colloidal-phase flushing agents would increase the aqueous concentrations of some hydrophobic organic contaminants in soils or aquifers by up to several hundred fold. Key words: humic, humalite, binding, hydrophobic, remediation

Introduction Aqueous humic substances bind hydrophobic contaminants, and thus exert control on their mobility and toxicity in the environment. This phenomenon is the basis for the current research into the potential use of commercial humic products as remediation products. Some investigators have examined the potential use of commercial humic products as sequestering agents (Shimizu 1998; Sanjay et al. 1999; Poerschmann and Balcke 2000). More often, researchers have considered the use of mobile, aqueous-phase humic substances as flushing agents to clean up contaminated soils and aquifers (Abdul et al. 1990; Xu et al. 1994; Ding and Wu 1997; Lesage et al. 1995; Rebhun et al. 1996; Johnson and John 1999; Boving and Brusseau 2000; Van Stempvoort et al. 2002; Molson et al. 2002). Most published data on the binding of organic contaminants by commercial aqueous humic products are studies with Aldrich humic acid (HA). The research with Aldrich HA, including studies over the past * Corresponding author; [email protected]

268

VAN STEMPVOORT ET AL.

8 years at our institute (Xu et al. 1994; Van Stempvoort and Lesage 2002), have focused on binding of PAHs, PCBs and pesticides (e.g., Burkhard 2000; Krop et al. 2001). Given the considerable attention that has been devoted to understanding the binding properties of Aldrich HA, it is not surprising that virtually all previous investigations into the use of commercial aqueous humic products (AHPs) as contaminant flushing agents have used Aldrich HA (see above references). Although some have reported nonlinear behaviour (e.g., Xu et al. 1994; Chiou et al. 2000; Yuan and Xing 2001; Laor and Rebhun 2002), in many studies the binding of organic contaminants to aqueous humic substances is modeled as linear partitioning: Koc = Cb/Cw

(1)

where Koc is the partitioning coefficient (L/g organic C), Cb is the humic-bound contaminant concentration (mass of bound contaminant, Mb, per unit concentration of humic organic carbon, Coc), and Cw is the dissolved contaminant concentration in water. We prefer to use the term Koc in equation 1, rather than Kdoc, which some authors (e.g., Burkhard 2000) have used to designate the humic-dominated aqueous organic matter as “dissolved organic carbon.” However, the bulk of aqueous humic substances form a polydisperse mixture of suspended colloids, between 1 nm and 1 µm in diameter. Such an aqueous colloidal dispersion can be referred to as a hydrosol (Olson et al. 1988). Researchers have found that Aldrich HA is generally a stronger binder of hydrophobic organic contaminants than most natural aqueous humic substances, whether derived from soils or surface waters (e.g., Chiou et al. 1987; Chin et al. 1997; Doll et al. 1999). Burkhard (2000) has recently compiled information on the binding of nonionic organic compounds to aqueous humic substances. He provided summary equations to predict the relationship between Koc (his Kdoc) to hydrophobicity, as measured by octanol-water partitioning. For Aldrich HA, the summary equation is: log(Koc) = 0.85(±0.03)*log(Kow) + 0.27(±0.20)

(2)

where Kow is the 1-octanol/water partitioning coefficient for the organic compound. For naturally occurring aqueous humic substances (“DOC”), the weaker binding is described by the following Burkhard equation: log(Koc) = 0.85(±0.06)*log(Kow) - 0.25(±0.34)

(3)

Although equations 2 and 3 have been presented by Burkhard as being applicable for hydrophobic organic contaminants in general, most of the experiments on which these equations are based were tests with aromatic compounds (see Results and Discussion).

BINDING OF ORGANICS TO HUMIC PRODUCTS

269

Most binding tests with Aldrich HA have been conducted at low humic concentrations. With Aldrich HA levels between 0.001 and 0.020 g OC/L (where OC = organic C), most studies have indicated that the binding of contaminants tends to decrease in strength (factor of ~2–3), measured as Koc, with increasing humic concentration (Carter and Suffet 1982; Landrum et al. 1984; Yin and Hassett 1986; Li et al. 1997). In contrast, for a similar range in Aldrich HA concentrations, Chiou et al. (1987) inferred relatively constant Koc, values. For all of these tests, the HA levels were much too low to be practical for efficient flushing in remediation applications. Studies at higher concentrations of Aldrich HA (>0.02 g OC/L) have given mixed results regarding the HA concentration effect on Koc values. Using a headspace technique, Resendes et al. (1992) found relatively constant Koc values for the binding of chlorobenzenes to Aldrich HA over an HA range up to ~0.3 g OC/L. Similarly, using batch solubilization and solid phase microextraction (SPME), respectively, Xu et al. (1994) and Doll et al. (1999) indicated relatively constant binding strength for phenanthrene to Aldrich HA over a combined range up to ~0.8 g OC/L HA. Doll et al. noted a slight decrease in binding strength with increasing HA. Using batch solubilization, Johnson and John (1999) inferred relatively constant binding strength for tetrachloroethene at Aldrich HA concentrations up to ~6 g OC/L. Using SPME, Van Stempvoort and Lesage (2002) found that the apparent Koc values for methylated naphthalenes (important PAHs in diesel and other petroleum products) decreased by a factor of ~2 to 3 when the Aldrich HA increased from 0.30 to 0.97 g OC/L, but were similar at 0.97 and 3.15 g OC/L. In notable contrast to the above studies, a batch solubilization study by Guetzloff and Rice (1994) suggested that purified Aldrich HA formed micelles at levels above ~3 to 4 g OC/L (they reported the critical level as 7.4 g/L total HA), and that the strength of the binding of the organic pesticide DDT was greatly enhanced above this critical micelle concentration (CMC). However, in a follow-up study, the same authors (1996) found no evidence for the same effect of the CMC on the binding of pyrene to Aldrich HA. Thus, there is not firm evidence for the existence of a consistent, predictable CMC for Aldrich HA. This study is a continuation of research at the National Water Research Institute (NWRI) in Canada on the potential use of AHPs in environmental remediation. Previous testing at the bench and pilot scale with Aldrich humic acid (HA) yielded promising results (Xu et al. 1994; Lesage et al. 1995; Van Stempvoort et al. 2000, 2002; Van Stempvoort and Lesage 2002). Given that Aldrich HA is a relatively expensive laboratory reagent, we decided to shift the focus of our research in order to investigate more affordable commercial humic products that might be used in full-scale field applications. In our search for commercial AHPs that can be used in full-scale contaminant remediation applications, the goal was to find one or more prod-

270


ucts that have contaminant-binding properties similar to those of Aldrich HA, but that are more cost-effective for use in field applications. This led to the current research initiative with AHPs provided by Luscar Ltd. of Edmonton, Alberta, Canada. The quotes that we obtained (2002) indicate that the current cost for bulk quantities of BlackEarth™, a dry, powder form, “soluble” humic product produced and marketed by Luscar Ltd., is about one-fifth of the cost for bulk, technical grade sodium-form Aldrich® HA (both per bulk weight basis). This report provides results of an investigation of the binding of volatile or semi-volatile organic contaminants to highly concentrated AHPs using a headspace technique. The advantages of this technique were that we could investigate the binding of contaminants to aqueous humic products at very high aqueous humic concentrations (up to 33.4 g OC/L), and we did not have to contend with matrix interference of the analyses, because we analyzed samples of the equilibrated headspace air, rather than the AHP hydrosols. We tested the binding of four chlorinated organic compounds: heptachlor, hexachlorobenzene, hexachloroethane and 1,2,4-trichlorobenzene, as well as a representative petroleum hydrocarbon, n-hexane. Because of their hydrophobic nature, these five compounds bind strongly to AHPs, and their moderate to high volatility made them suitable for GC analyses using the headspace technique. The four chlorinated compounds are toxic and persistent organic pollutants (POPs) of global concern. Heptachlor and hexachlorobenzene, formerly in wide use as pesticides, are 2 of the 12 POPs targeted to be banned under the international Stockholm Convention on Persistent Organic Pollutants (May 22, 2001).

Theory and Equations for the Headspace Technique Headspace sampling is commonly used to measure the concentrations of volatile organic compounds in water. Theory underlining practical applications of the headspace technique to examine the binding of contaminants to aqueous humic substances have been provided by Yin and Hassett (1986), Resendes et al. (1992) and others. The specific procedures and equations that we used at NWRI pertain to sampling the headspaces of static batches (see Materials and Methods section). A basic assumption of all headspace techniques is Henry’s law, which describes the equilibrium partitioning behaviour of a solute between two solvent phases, in this case a (semi-)volatile organic contaminant between water and air. In the case of static batch tests with dilute solutions, the activity of each solute in each phase is assumed to be proportional to the concentration of the solute in that phase and independent of the presence of other solutes (i.e., activity coefficients = 1). Assuming ideal behaviour in dilute solutions at a constant temperature, this relationship can be expressed as follows:


271

H = Ca/Cw

(4)

where H is the dimensionless Henry’s law coefficient, and Ca and Cw are the concentrations of the contaminant in the headspace (air) and water, respectively. In static batch tests, Ca can be defined as Ma/Va, and Cw as Mw/Vw, where Ma is the mass in the headspace, Va the headspace volume, Mw the mass in the water, and Vw the volume of water. In the headspace technique, Ca is analyzed, generally by gas chromatography (GC). Using the above relationships: Ca = H*Mt/(Vw + H*Va)

(5)

where Mt is the total mass of contaminant added, the sum Ma + Mw. If reliable values of H are not available, they can be determined in paired batch tests, “1” and “2”, by varying the ratio of Va to Vw and applying the Ca measured for each (Lincoff and Gossett 1984): H = [(Ca1/Ca2)*Vw1-Vw2]/[Va2-(Ca1/Ca2)Va1]

(6)

The use of equation 6 assumes a linear relationship between the detector response for the analysis (e.g., peak area for GC analyses) and Ca. For headspace batch tests with AHPs, the binding of the contaminant to humic substances (equation 1) also has to be taken into consideration. The volatile or semi-volatile organic contaminant partitions between three phases, air, water (free dissolved) and the aqueous humic product (bound). In this case, Mt = Ma + Mw + Mb, and the following relationship holds: Ca = {(H*Mt)/[Vw*(1 + X)]}/{(1 + H*Va)/[Vw*(1 + X)]}

(7)

where X = Koc*Coc. In tests with high concentrations of AHPs (up to several weight %), the activity of the water is lower, by perhaps several percent compared to tests with only water. Consequently, the use of equation 7, which assumes ideal behaviour, results in small errors. Lower water activity in the presence of the AHPs is numerically equivalent to decreasing the value of Vw in equation 7. Accordingly, the calculated Koc values reported in this paper, which are based on equation 7, may underestimate this binding coefficient by up to several percent at the higher AHP concentrations.

Materials and Methods Aqueous humic products (AHPs) L-155 and L-160 were provided by Luscar Ltd. (1600 Oxford Tower, 10235-101 Street, Edmonton, Alberta, Canada T5J 3G1). These Luscar test products are derived from humalite deposits in Alberta, Canada. Humalite, which is similar to leonardite, is a

272


naturally occurring, highly oxidized organic material, found adjacent to sub-bituminous coal, with high concentrations of humic acids (Hoffman et al. 1993). Recently, Schnitzer et al. (2001) and Dinel et al. (2001) have characterized the chemistry of similar Luscar humic products derived from the same humalite deposits in Alberta. Concentrated Aldrich HA was obtained for comparative testing, by adding between 13.25 and 99.38 g of the sodium salt (tech., product no. H16752, Lot No. 16206AN, Aldrich Chemicals, Milwaukee, Wis.) per L of Milli-Q water. When portions of the Aldrich HA hydrosols were taken for batch testing, the precipitate fractions (~20% of HA) were excluded. The AHPs were diluted in Milli-Q water as required, and their total organic carbon measured with a Shimadzu TOC-5050 analyzer. The organic chemicals to be tested were obtained as reagent-grade heptachlor, 1,2,4-trichlorobenzene (TCB) and n-hexane from Sigma Aldrich Canada Ltd. (Oakville, Ontario), hexachlorobenzene (HCB) and hexachloroethane (HCA) from Supelco (Oakville, Ontario). Stock solutions were prepared in methanol (1 or 5 ppm HCA with 100 or 500 ppm TCB; 1167 ppm n-hexane) or 50/50 ether and methanol (HCB with heptachlor, 5 or 100 ppm each). These stock solutions were stored at 5°C in the dark in 4-mL glass vials sealed with teflon-lined septa, diluted in methanol and/or mixed as required. For each batch test, a small quantity (1.5 to 100 µL) of a stock solution or a derivative thereof was transferred by microsyringe through a teflon-lined butyl rubber septum into a 60 mL (160 mL for n-hexane) glass serum bottle containing either Milli-Q water or an AHP (full strength or diluted), with a headspace. In preliminary tests, batches placed on an orbital shaker at ~100 rpm (New Brunswick Scientific, G-33) required approximately 48 hours to reach steady Ca values, inferred to indicate equilibrium between aqueous (dissolved, humicbound) and headspace phases for the contaminants. All batch tests were conducted at 23 ± 2°C. For each batch, the volume of water or AHP was determined gravimetrically, using density corrections for AHPs. Headspace volumes were determined gravimetrically by topping up each batch with Milli-Q water after analysis. Volumes of water or AHPs for the batch tests varied between 3 and 55 mL, depending on the volatility of the contaminant tested. For the tests with water for determination of H values, and for later tests with AHPs, at least 2 different ratios of Va/Vw were used. For each equilibrated batch, 100 µL of headspace air was sampled by Hamilton gastight microsyringe and injected into an SRI 8610A gas chromatograph, equipped with a DB-1, DB-5 or DB 624 column. For the chlorinated compounds, Henry’s law coefficients were determined using equation 6; for n-hexane, the Henry’s coefficient was based on a literature value (see Results and Discussion). Tests with Milli-Q water were used to calibrate the Ca in tests with AHPs. In the latter tests, we applied equation 7, using iteration to solve X, matching calculated and measured Ca, as analyzed by GC. With this X value, Koc could be calculated, since Coc was known.


273

Results and Discussion Henry’s Law Coefficients As determined in headspace batch experiments with Milli-Q water, Henry’s law coefficients for the four organochlorine contaminants are shown in Table 1. These coefficients are within the ranges previously reported (Mackay et al. 1992; Sander 1999). Our values for hexachloroethane (HCA) and heptachlor are within 5% of values in the summary tables of Mackay et al. (1992, Vol. III and V). The value obtained for TCB is approximately 25% higher than a recent value reported by Dewulf and Van Langenhove (2001). The published values for HCB vary by 4 orders of magnitude (Sander 1999). For n-hexane, we chose a literature H value of 74. The majority of the values compiled by Makay et al. (1992, Vol. III) and Sander (1999) are within 15% of this value. Contaminant Binding to AHPs For the five contaminants, the masses of each added to the batches, and the resulting free dissolved concentrations were similar, except for HCA (Table 2). Given the relatively high volatility of HCA, as indicated by Table 1, and its very high response factor on the electron capture detector, less of this compound was added to each batch in order to fit the range of linear responses for the detector. The headspace concentration of n-hexane was much higher than for the other contaminants, because of its very high volatility, and the much different response of the flame ionization detector. The partitioning and binding of heptachlor and HCB were very similar (Table 2). For comparable ranges of dissolved TCB and hexane, the binding of these contaminants to L-155 was similar (Table 2). The contaminant binding coefficients, Koc, obtained for the batch tests with L-155, L-160, and Aldrich HA, are shown in Tables 3, 4 and 5, respectively. More data was acquired for L-155 than the other AHPs. For the majority of the Koc data, standard deviations are within 25% of the means. The Koc for each contaminant is relatively consistent over the ranges in humic concentrations tested. These results suggest that the H values that we used were accurate. Errors in H values would be incorpo-

Table 1. Henry’s law coefficients (dimensionless, see equation 4)

Compound hexachloroethane hexachlorobenzene 1,2,4-trichlorobenzene heptachlor

µ ± σ (n) 0.34 ± 0.03 (16) 0.25 ± 0.06 (18) 0.20 ± 0.04 (12) 0.15 ± 0.05 (9)

0.09–1.4 0.26–1.7 0.9–8.3 0.01–0.25 492–2977

µg/L (Ca)

µg (Mt)

1.5–10.7 1.5–10.7 0.5–3.1 0.005–0.031 18.5

Headspace conc.

Total mass in batch

0.6–6.3 1.1–6.8 4.4–41.4 0.08–0.3 11.5–40.2

µg/L (Cw)

Dissolved conc.

10.5–88.8 10.5–88.7 2.6–25.5 0.01–0.10 3.16–10.6

µg/g organic C (Mb/Coc)

Humic-bound conc.

51.3–1051 51.1–1048 14.9–287 0.13–2.5 38.1–110

µg/L (Cw + Mb/Vw)

Total aqueous conc.

heptachlor hexachlorobenzene 1,2,4-trichlorobenzene hexachloroethane n-hexane

Humic organic C conc. (g/L)

16.4 ± 2.6 (8) 17.0 ± 3.5 (10) 0.63 ± 0.20 (9) 0.41 ± 0.19 (13)

1.67 18.3 ± 3.4 (6) 17.4 ± 2.0 (6) 0.60 ± 0.12 (6) not det. 0.26 ± 0.03 (6)

3.34

15.2 ± 2.1 (8) 11.8 ± 1.4 (7) 0.79 ± 0.16 (5) 0.14 ± 0.04 (5)

6.68

11.4 ± 4.0 (6) 16.0 ± 4.9 (6) 0.32 ± 0.05 (6) 0.27 ± 0.04 (6) 0.26 ± 0.02 (6)

16.7

14.5 ± 2.3 (6) 8.16 ± 1.63 (6) 0.29 ± 0.07 (6) 0.26 ± 0.02 (6)

33.4

Table 3. Koc (L/g) values, µ ± σ (n), measured in this study for the binding of various organic contaminants (left column) to Luscar product L-155, at 5 humic concentrations

heptachlor hexachlorobenzene 1,2,4-trichlorobenzene hexachloroethane n-hexane

Contaminant

Table 2. Ranges in results for equilibrated batch tests with selected contaminants and Luscar aqueous humic product L-155

274 VAN STEMPVOORT ET AL.

275


Table 4. Koc (L/g) values, µ ± σ (n), measured in this study for the binding of various organic contaminants to Luscar product L-160, at 2 humic concentrations


11.6

heptachlor hexachlorobenzene 1,2,4-trichlorobenzene hexachloroethane

29.0

13.0 ± 1.0 (6) 15.2 ± 0.9 (6) 0.46 ± 0.09 (5) 0.19 ± 0.02 (5)

12.8 ± 2.8 (6) 13.4 ± 2.7 (6) 0.38 ± 0.04 (5) 0.18 ± 0.02 (5)

rated into calculations by equation 7, enlarging the variances of the resulting Koc over the Coc, Va and Vw ranges that were tested. The Koc values for the 4 organochlorine contaminants were similar in tests with the two Luscar products L-155 (Table 3) and L-160 (Table 4). The data for L-155 were obtained over a larger range in humic concentrations. For similar ranges in humic C concentrations (≥ 6.68 g OC/L), there were no significant differences in the Koc values for each contaminant in L-155 and L-160, based on t-tests (2-tail, 0.05 level). Similar t-tests were used to compare results for the largest differences in L-155 concentrations, 1.67 and 33.4 g OC/L (Table 3). These indicated that the Koc values for HCB and TCB at 33.4 g OC/L were significantly lower than those at 1.67 g OC/L. In contrast, there was no significant difference in the Koc values for heptachlor and HCA at these two humic concentrations. Based on t-tests as above, the marginally higher values of Koc for heptachlor and HCA in Aldrich HA with an increase in HA from 9.34 to

Table 5. Koc (L/g) values for binding of various organic contaminants to Aldrich humic acid

Measured in previous studiesa heptachlor hexachlorobenzene 1,2,4-trichlorobenzene hexachloroethane

30.2 93, 138, 269 1.29, 4.79, 10, 13


not reporteda

a Supporting

Measured in this study, µ ± σ (n) 42.0 ± 8.1 (5) 49.5 ± 6.8 (5) 2.58 ± 1.02 (5) 1.05 ± 0.10 (5)

information in Burkhard (2000).

9.34

57.5 ± 6.0 (5) 49.5 ± 6.7 (5) 2.51 ± 0.35 (5) 1.55 ± 0.10 (5) 21.7

276


21.7 g OC/L are significant. In contrast, there were no significant changes in the Koc values for HCB and TCB over this range in Aldrich HA. The data shown in Tables 3 through 5 provide no support for an earlier conclusion that, at high concentrations (above a CMC), aqueous humic substances form micelles, resulting in sharp increases in Koc values (Guetzloff and Rice 1994). In contrast, our results at elevated humic concentrations indicate small increases or decreases in Koc, or negligible changes. Our results are consistent with the findings of a number of previous studies for Aldrich HA (Xu et al. 1994; Doll et al. 1999; Johnson and John 1999; Van Stempvoort and Lesage 2002) and other AHPs (Kurz et al. 1998). A comparison of Tables 3 through 5 indicates that Aldrich HA is the strongest binder of the organic contaminants. For the four chlorinated compounds tested, the Koc values for Aldrich are approximately 3 to 7 times larger than those for the two Luscar AHPs, L-155 and L-160. For HCB and heptachlor, the Koc values for binding to L-155 and L-160 are close to those predicted by equation 3 (Fig. 1), typically varying from the predicted values by factors of 0.65 to 1.1. Equation 3 (Burkhard 2000) summarizes the binding behaviour of naturally occurring aqueous humic substances. Thus, our data suggest that the binding of these contaminants by the Luscar AHPs was similar in strength to the binding that occurs by typical humic substances that occur in natural aquatic environments. However, the Koc values for TCB, HCA and n-hexane to L-155 were less closely predicted by equation 3 (Fig. 1), typically lower by factors of 0.11 to 0.37. A similar pattern was found when our data for Aldrich HA was compared to equation 2, which is based on a compilation of previous data for the binding of a large range of contaminants to Aldrich HA (Burkhard 2000). The Koc values for the binding of heptachlor and HCB to Aldrich HA (Table 4) fall close to the predicted values based on equation 2 (Fig. 1), varying from them by factors of 0.56 to 1. The measured Koc values for the binding of TCB and HCA are smaller than those predicted by equation 2 (Fig. 1), by factors of 0.26 to 0.45. The above results indicate that, of the five compounds tested, the binding of heptachlor and of HCB match most closely the Burkhard equations. This interpretation is dependent on the H and Kow values that we selected for the contaminants. The data used by Burkhard to generate equations 2 and 3 were largely based on tests with compounds dominated by aromatic C (PAHs, PCBs, pesticides such as DDT), and only a few tests for aliphatic-C dominated compounds were included. As such, it appears that the Burkhard equations should not be used to predict the binding of aliphatic compounds, such as HCA and n-hexane. Such an interpretation is consistent with the conclusions of Goss and Schwarzenbach (2001), Poerschmann and Kopinke (2001) and others, who conclude that empirical equations that relate Koc to Kow are not be widely applicable to different groups of nonionic organic compounds. Each equation developed to summarize or predict Koc-Kow relationships should be based on tests with a single, defined class of organic compounds that have similar/common chemical structures and properties.


277

Fig. 1. Graphical comparison of the binding coefficients (Koc) obtained in this study compared to those predicted by equations 2 and 3 in the text (from Burkhard 2000). For the compounds studied, selected Kow values as tabulated by Mackay et al. (1992) were used.

Despite its superior binding strength, Aldrich HA is an expensive product and thus it will probably not be used as a commercial remediation product in large-scale field applications. By comparison, if other commercial products, such as the humalite-derived Luscar AHPs tested, are marketed as relatively inexpensive alternatives, they may prove to be useful as commercial remediation products. As shown in Fig. 2, the potential increases in aqueous concentrations of some organic contaminants during flushing of contaminated soils or aquifers could be up to several hundred fold in the presence of concentrated Luscar AHPs added as flushing agents. Thus, these AHPs might prove to be useful flushing agents to speed the cleanup of these soils and aquifers. Based on this study, the two Luscar AHPs, L-155 and L-160, appear to be better proxies for typical natural aqueous humic substances than

278


Fig. 2. Potential enhancement of the total aqueous concentration ([bound + dissolved]/dissolved) of the contaminants in a soil/aquifer flushing application with Luscar product L-155, based on the Koc data shown in Table 3.

Aldrich HA. This observation suggests that it may be possible to develop one or more Luscar products specifically for use as a laboratory standard reference material for testing the binding of organic contaminants to representative aqueous humic substances. Such a commercial product might provide a useful alternative to Aldrich HA, for which a large volume of data has already been obtained. It would be useful to conduct further studies to compare the chemical structures and/or sorbant properties of Luscar AHP to those of natural aqueous humic substances. Our data indicate that equation 3 (Burkhard 2000) may be useful to predict the binding of strongly hydrophobic chlorinated aromatic compounds to Luscar AHPs, such as heptachlor and HCB. Further work is required to confirm this for a larger range of Luscar products, and/or to determine other suitable parameters and/or equations for the binding of various groups of organic contaminants (e.g., chlorinated and nonchlorinated, aromatics and aliphatics) to Luscar AHPs.

Conclusions The results of this study indicate that there are minor or negligible changes in binding strength over the full range of aqueous concentrations tested for two humalite-derived humic products provided by Luscar Ltd.,


279

and for Aldrich humic acid (HA). This is consistent with other findings for Aldrich HA and other humic products. The data provide no support for earlier reports that, at high concentrations, aqueous humic substances form micelles, resulting in sharp increases in Koc values. Aldrich HA was a stronger binder of the five hydrophobic organic compounds tested than the two Luscar humic products, by factors of approximately 3 to 7. Our data indicate that the binding of the selected hydrophobic organics to the two Luscar products is similar to the binding predicted for humic substances that occur in natural aquatic environments. It may be useful to develop one or more Luscar products specifically for use as a laboratory standard reference material for testing the binding of organic contaminants, as an alternative to Aldrich HA, for which a large volume of data has already been obtained. Pending further study, Luscar humic products might better simulate natural organic matter for many aquatic environments. Our binding data suggest that increases in total aqueous concentrations of some organic contaminants in soils or aquifers could approach several hundred fold in the presence of concentrated Luscar humic products added as flushing agents. Thus, these products might prove to be useful as commercial flushing agents for soil and groundwater remediation.

Acknowledgements This study was funded by Luscar Ltd. (Edmonton, Alberta, Canada) and Environment Canada.

References Abdul AS, Gibson TL, Rai DN. 1990. Use of humic acid solution to remove organic contaminants from hydrogeologic systems. Environ. Sci. Technol. 24:328–333. Boving TB, Brusseau ML. 2000. Solubilization and removal of residual trichloroethene from porous media: comparison of several solubilization agents. J. Contam. Hydrol. 42:51–67. Burkhard LP. 2000. Estimating dissolved organic carbon partition coefficients for nonionic organic chemicals. Environ. Sci. Technol. 34:4663–4668. Carter CW, Suffet IH. 1982. Binding of DDT to dissolved humic materials. Environ. Sci. Technol. 16:735–740. Chin Y-P, Aiken GR, Danielsen KM. 1997. Binding of pyrene to aquatic and commercial humic substances: the role of molecular weight and aromaticity. Environ. Sci. Technol. 31:1630–1635. Chiou CT, Kile DE, Brinton TI, Malcolm RL, Leenheer JA. 1987. A comparison of the water solubility enhancements of organic solutes by aquatic humic materials and commercial humic acids. Environ. Sci. Technol. 21:1231–1234. Chiou CT, Kile DE, Rutherford DW, Sheng G, Boyd SA. 2000. Sorption of selected organic compounds from water to a peat soil and its humic-acid and humin fractions: potential sources of the sorption nonlinearity. Environ. Sci. Technol. 34:1254–1258.

280


Dewulf J, Van Langenhove H. 2001. Solid-phase microextraction of volatile organic compounds in environmental applications. American Laboratory 33:18–20. Dinel H, Schnitzer M, Paré T, Schulten H-R, Ozdoba D, Marche T. 2001. Interpretation by principle-component analysis of pyrolysis-field ionization mass spectra of lignite ores, p. 329–336. In Ghabbour EA, Davies G (ed.), Humic substances: structures, models and functions. Royal Society of Chemistry, Cambridge, UK. Ding J-Y, Wu S-C. 1997. Transport of organochlorine pesticides in soil columns enhanced by dissolved organic carbon. Water Sci. Technol. 35:139–145. Doll TE, Frimmel FH, Kumke MU, Ohlenbusch G. 1999. Interaction between natural organic matter (NOM) and polycyclic aromatic compounds (PAC) – comparison of fluorescence quenching and solid phase micro extraction (SPME). Fresenius J. Anal. Chem. 362:313–319. Goss K-U, Schwarzenbach RP. 2001. Linear free energy relationships used to evaluate equilibrium partitioning of organic compounds. Environ. Sci. Technol. 35:1–9. Guetzloff TF, Rice JA. 1994. Does humic adic form a micelle? Sci. Tot. Environ. 152:31–35. Guetzloff TF, Rice JA. 1996. Micellar nature of humic colloids, p. 18–25. In Gaffney JS, Marley NA, Clark SB (ed.), Humic and fulvic acids, isolation, structure and environmental role. ACS Symposium Series 651, American Chemical Society, Washington, DC. Hoffman GL, Nikols DJ, Stuhec S, Wilson RA. 1993. Evaluation of leonardite (humalite) resources of Alberta. Open File Report 93–18, Alberta Geological Survey, Edmonton, Alta. Prepared by Retread Resources Ltd. (Calgary, Alta.) for Energy, Mines and Resources Canada. Johnson WP, John WW. 1999. PCE solubilization by commercial humic acid. J. Contam. Hydrol. 35:343–362. Krop HB, van Noort PCM, Govers HAJ. 2001. Determination and theoretical aspects of the equilibrium between dissolved organic matter and hydrophobic organic micropollutants in water (Kdoc). Rev. Environ. Contam. Toxicol. 69:1–122. Kurz MD, Olson ES, Gallagher JR. 1998. Task 1.16 – Enhanced mobility of dense nonaqueous-phase liquids (DNAPLs) using dissolved humic acids, 98EERC-10-03, Energy and Environmental Research Center, University of North Dakota, Grand Forks, N. Dak. Prepared for U.S. Dept. Energy, Federal Energy Technology Center, Pittsburg, Pa. Landrum PF, Nihart SR, Eadie BJ, Gardner WS. 1984. Reverse-phase separation method for determining pollutant binding to Aldrich humic acid and dissolved organic carbon of natural waters. Environ. Sci. Technol. 18:187–192. Laor Y, Rebhun M. 2002. Evidence for nonlinear binding of PAHs to dissolved humic acids. Environ. Sci. Technol. 36:955–961. Lesage S, Novakowski KS, Xu H, Bickerton G, Durham L, Brown S. 1995. A large scale aquifer model to study the removal of aromatic hydrocarbons from the saturated zone, 6 p. In Proceedings of Solutions’95, International Association of Hydrogeologists Congress, Edmonton, Alberta. Li AZ, Marx KA, Walker J, Kaplan DL. 1997. Trinitrotoluene and metabolites binding to humic acid. Environ. Sci. Technol. 31:584–589. Lincoff AH, Gossett JM. 1984. The determination of Henry’s constant for volatile organics by equilibrium partitioning in closed systems, p. 17–25. In Brutsaert W, Jirka GH (ed.), Gas transfer at water surfaces. D. Reidel Publishing Co., Dordrecht, Holland.


281

Mackay D, Shiu WY, Ma KC. 1992. Illustrated handbook of physical-chemical properties and environmental fate for organic chemicals. Volumes I, III and V, Lewis Publishers, Chelsea, Mich. Molson JW, Frind EO, Van Stempvoort DR, Lesage S. 2002. Humic acid enhanced remediation of an emplaced diesel source in groundwater: 2. Numerical model development and application. J. Contam. Hydrol. 54:77–305. Olson ES, Diehl JW, Froelich ML. 1988. Hydrosols from low rank coals. 1. Preparation and properties. Fuel 67:1053–1061. Poerschmann J, Balcke G. 2000. Application of humic acids in in-situ groundwater remediation. Program with Abstracts, Humic Substances Seminar IV, March 22–24, 2000, Northeastern University, Boston, Mass. Poerschmann J, Kopinke F-D. 2001. Sorption of very hydrophobic organic compounds (VHOCs) on dissolved humic organic matter (DOM). 2. Measurement of sorption and application of a Flory-Huggins concept to interpret the data. Environ. Sci. Technol. 35:1142–1148. Rebhun M, de Smet F, Rwetabula J. 1996. Dissolved humic substances for remediation of sites contaminated by organic pollutants. Water Res. 30:2027–2038. Resendes J, Shiu WY, Mackay D. 1992. Sensing the fugacity of hydrophobic organic chemicals in aqueous systems. Environ. Sci. Technol. 26:2381–2387. Sander R. 1999. Compilation of Henry’s Law constants for inorganic and organic species of potential importance in environmental chemistry. http://www. mpch-mainz.mpg.de/~sander/res/henry.html (Version 3, April 8, 1999). Sanjay HG, Walia D, Fataftah A. 1999. A new multipurpose remediation media. Environ. Protection 10:42–45,50. Schnitzer M, Dinel H, Paré T, Schulten H-R, Ozdoba D. 2001. Some chemical and spectroscopic characteristics of six organic ores, p. 315–328. In Ghabbour EA, Davies G (ed.), Humic substances: structures, models and functions. Royal Society of Chemistry, Cambridge, UK. Shimizu Y, Sogabe H, Terashima Y. 1998. The effects of colloidal humic substances on the movement of non-ionic hydrophobic organic contaminants in groundwater. Water Sci. Technol. 38:159–167. Van Stempvoort DR, Lesage S. 2002. Binding of methylated naphthalenes to concentrated aqueous humic acid. Adv. Environ. Res. 6:495–504. Van Stempvoort DR, Lesage S, Novakowski KS, Millar K, Brown S, Lawrence JR. 2002. Humic acid enhanced remediation of an emplaced diesel source in groundwater: 1. Laboratory-based pilot scale test. J. Contam. Hydrol. 54:249–276. Van Stempvoort DR, Molson JW, Lesage S, Brown S. 2000. Sorption of Aldrich humic acid to a test aquifer material and implications for subsurface remediation, p. 153–163. In Ghabbour EA, Davies G (ed.), Humic substances: versatile components of plants, soils and water. Royal Society of Chemistry, Cambridge, UK. Xu H, Lesage S, Durham L. 1994. The use of humic acids to enhance removal of aromatic hydrocarbons from contaminated aquifers, p. 635–645. In Proceedings, 4th Annual Symposium on Groundwater and Soil Remediation, Calgary, Alta. Yin C, Hassett JP. 1986. Gas-partitioning approach for laboratory and field studies of mirex fugacity in water. Environ. Sci. Technol. 20:1213–1217. Yuan G, Xing B. 2001. Effects of metal cations on sorption and desorption of organic compounds in humic acids. Soil Sci. 166:107–115.


An Assessment of Long-Term Monitoring Data for Constructed Wetlands for Urban Highway Runoff Control AARON C. FARRELL* AND RONALD B. SCHECKENBERGER Philips Engineering Limited, Burlington, Ontario

Constructed wetlands have gained acceptance as a means of treating stormwater runoff from urban developments. Much of the available data regarding the performance of these facilities is based upon monitoring conducted over the course of less than two years, and as such inherently assumes that the period of analysis represents the “typical” or “design” conditions under which these facilities are intended to operate. While this information has provided guidance regarding the mechanisms by which wetlands provide quality treatment of urban runoff, it does not fully reflect the variability of conditions under which the facilities operate over the fullness of time, which is of particular concern to designers and operators. The construction of the Dartnall Road Interchange, as part of Hamilton’s Lincoln Alexander Parkway, required a monitoring program—which included five years of water quality sampling—as a condition of approval by the Department of Fisheries and Oceans. This paper reports on the quantitative and qualitative wetland water quality monitoring data (sediment, nutrients, metals) obtained over the course of a total seven-year program, and provides information regarding the operating conditions and estimates on contaminant removal efficiencies from the facilities. Key words: wetland, stormwater management, monitoring

Introduction The Dartnall Road Interchange is located within the Red Hill Creek watershed in the City of Hamilton, Ontario, Canada (Fig. 1). The interchange connects Dartnall Road with the Lincoln Alexander Parkway. Stormwater quality control for runoff from the interchange ramps and a portion of the Lincoln Alexander Parkway is provided by three constructed wetland facilities. The wetlands were constructed in 1996 and became fully operational in 1997 when the Lincoln Alexander Parkway opened to vehicular traffic. Investigations by fisheries specialists prior to construction, determined the presence of fish species within the Red Hill Creek which flows through the interchange (Philips Planning and Engineering Ltd. 1993). Given that the interchange could not be relocated on account of the transportation needs, the construction of the interchange in the planned loca* Corresponding author; [email protected]

284

FARRELL AND SCHECKENBERGER

Fig. 1. Red Hill Creek watershed.

tion constituted a harmful alteration, disturbance or destruction (HADD) to fish habitat, hence approval was required from the federal Department of Fisheries and Oceans (DFO). As a condition of DFO permitting, a seven-year monitoring program, which was to include water quality sampling, was prepared and submitted for agency approval (Scheckenberger et al. 1996). The monitoring program included water quality analyses of runoff at the inlets to, and outlets from, the wetlands. Water quality sampling commenced in 1997, and in 1998 a sampling protocol was established in order to evaluate the removal efficiencies of the facilities. The water quality sampling continued until the completion of the field monitoring program in the fall of 2001. The monitoring program assessed the operating conditions and effectiveness of wetlands over an extended period, beyond that which is typically completed for such systems (i.e., 12 months or less), and has facilitated an assessment of wetland performance which accounts for

LONG-TERM MONITORING FOR CONSTRUCTED WETLANDS

285

seasonal variations in operating conditions which occur from one year to the next. This paper highlights the findings from this monitoring program and provides conclusions and recommendations for future studies.

Wetland Facility Designs The three subject wetlands provide quality treatment of highway stormwater runoff at the Dartnall Road Interchange (Fig. 2). At the time of design, the Red Hill Creek represented a warm-water fisheries habitat. The governing criteria of the day required a permanent storage volume roughly equal to the volume generated by 13 mm of rainfall over the entire drainage area, with an equal storage volume within the “intermittently flooded” (i.e., extended detention) portion of the facility above the normal water surface elevation (MOE/MNR 1991). It was further required that the volume within the extended detention portion of the facility be detained for no less than 24 hours. Table 1 summarizes the design parameters of each wetland facility (Philips Planning and Engineering Ltd. 1993). Each facility receives runoff from the highway surfaces. In addition, Wetland 3 receives runoff from approximately 15 ha of a nearby residential development. Wetland 1 and Wetland 2 have been designed to operate in series (Fig. 2). The 5.01-ha drainage area to Wetland 1 has been included in the design of the permanent pool and extended detention volumes for Wetland 2. Therefore, the combination of Wetland 1 and Wetland 2 provides approximately twice the requisite volume to service the 5.01-ha drainage area.

Fig. 2. Water quality monitoring stations at Dartnall Road Interchange.

286


Table 1. Summary of wetland design criteria

Wetland no. 1 2 3

Drainage area (ha)

Impervious coverage (%)

Permanent pool volume (m3)

Extended detention volume (m3)

Maximum controlled discharge (L/s)

5.01 5.84a 23.02 (7.92)b

35.9 35.3 28.5 (39.0)

660 737 1050

800 813 1050

9.45 9.45 12.10

a

Drainage area to Wetland 2 includes drainage area to upstream Wetland 1. Values in parentheses represent the proportion of the total drainage area which is highway drainage only. b

The Water Quality Monitoring Program Water Quality Sampling Methodology The water quality component of the monitoring program essentially extended from May 1 to October 31 for the years 1997 to 2001 inclusive. Water quality samples were obtained for three storm events for each year of the monitoring program. It was required that the highway surfaces have sufficient time to accumulate contaminants, and the storm intensity be sufficient to remove the contaminants from the highway surfaces. The events used for sampling were typically based upon a qualitative or predictive assessment of the storm’s intensity at the time of the event and shortly before sampling commenced, as well as the period between storms (i.e., the inter-event period) during which contaminants would accumulate on the highway surfaces. Water quality grab samples were taken at each inlet to the wetlands (Stations 2, 4 and 5 of Fig. 2), as well as within the Red Hill Creek proper upstream and downstream of the Dartnall Road Interchange (Stations 1 and 7 of Fig. 2), during or shortly after each storm event. In order to obtain a representative or “average” sample of the water quality during the first flush, a composite sample was obtained at each station by half-filling the containers during the first half of the storm (i.e., on the rising limb of the hydrograph) and topping-off the containers during the second half of the storm (i.e., on the receding limb of the hydrograph). Water quality grab samples were taken at the outlet of Wetland 2 and Wetland 3 (Stations 3 and 6 of Fig. 2) approximately 12 hours after the onset of the storm event and the initial “first-flush” sampling. This delay typically allowed time for the stormwater runoff to travel through the wetlands to the outlet, and allowed the creek level to fall below the level of the wetland outlet, thereby avoiding any mixing of the wetland efflu-

287


ent and creek water. This protocol for obtaining water quality samples at the wetland outlets was established for the last four years of the monitoring program. All water quality grab samples were analyzed by a licensed laboratory for pH, BOD5, anions, nutrients, microbiology and metals. This provided a more comprehensive basis for evaluating the loading conditions under which the wetlands operate, as well as determining removal efficiencies of these facilities. Sampling Event Characteristics Precipitation data for each of the storms monitored were obtained from the Atmospheric Environment Service rain gauge in the City of Hamilton at Mount Hope, located approximately 9 km southwest of the Dartnall Road Interchange. Given that the drainage area and the rain gauge are both at the same general elevation, this was considered an appropriate gauge to represent rainfall input. Characteristics of each storm sampled are provided in Table 2. The data in Table 2 indicate that over the course of the five years of water quality monitoring, the wetlands have been assessed for a variety of rainfall distributions and intensities. As well, the wetlands have been evaluated for events during abnormally dry summers (such as 1999 and 2001) as well as wet summers (such as 2000), which consequently influTable 2. Summary of storm events for monitoring season

Monitoring year

Storm event

Volume (mm)

Duration (hours)

Inter-event period (days)

1997

August 27 September 10 October 26

14.4 6.2 38.0

0.75 3.75 16.0

3 3 5

1998

June 30 August 18 October 7

14 4 8.8

2.0 2.25 7.0

5 3 6

1999

July 24 September 29 October 13

7.7 21.1 35.6

2.25 8.17 10.32

5 5 5

2000

May 12 June 2 September 20

7.2 9.8 3.8

1.75 0.75 1.0

2 8 5

2001

August 20 September 20 October 23

1.4 5.2 4.6

0.33 3.0 0.5

1 15 6

288


enced the loading conditions (i.e., contaminant concentrations) to the wetlands, despite comparable inter-event periods. Water Quality Data The grab samples at the Dartnall Road Interchange have been analyzed for 46 specific contaminants and water quality indicators, including five-day biochemical oxygen demand (BOD5), total suspended solids (TSS), total Kjeldahl nitrogen (TKN), total phosphorus (Total P), fecal streptococci (Fec. Strep.), Escherichia coli (E. coli), copper (Cu), iron (Fe), lead (Pb), and zinc (Zn). The contaminant concentrations at the inlets of Wetland 2 and Wetland 3 are presented in Table 3 for the events sampled from 1997 to 2001 inclusive (Philips Engineering Ltd. 1996–2001). As noted earlier, Wetland 1 and Wetland 2 receive runoff exclusively from highway surfaces, at comparable, if not identical, levels of impervious coverage. As such, the contaminant concentrations to Wetland 2 are representative of the contaminant concentrations entering Wetland 1. (Note: The gaps in the water quality data have resulted from occasions where inflow to, and/or discharge from, the wetland was not observed during sampling, as well as errors on the part of the analyzing lab when handling or storing the samples.) The contaminant concentrations at the inlet of Wetland 2 and Wetland 3 have been analyzed in order to determine “typical” contaminant concentrations from highway surfaces. The mean and median concentrations for each contaminant at the wetland inlets are compared in Table 4, and Fig. 3 and 4. (Note: All statistical analyses of contaminant concentrations have been completed using a lognormal distribution.) The observed contaminant concentrations within the Dartnall Road wetlands indicate that the concentrations of BOD5, TKN, Pb, and Zn were comparable at the inlets of Wetland 2 and Wetland 3. The mean concentration for Fe at the inlet to Wetland 3 was greater than that observed for Wetland 2; this result is anomalous, given the land use draining to each inlet. However, the median concentrations are comparable between the two facilities, and are both comparable to the mean concentration at the inlet to Wetland 3. This indicates that the mean concentrations of Fe at the inlet to Wetland 2 were likely skewed by anomalous concentrations of Fe obtained during the monitoring program. The mean and median concentrations of Cu at the inlet to Wetland 3 are greater than the concentrations observed at the inlet to Wetland 2; this is supported by the concentration data presented in Table 3. Although this result is unexpected given the land use draining to each facility, the consistency of this result suggests a higher traffic density west of the interchange, which would contribute to the contaminant concentrations entering Wetland 3. A literature review has been completed in order to verify the range of contaminant concentrations observed for each facility, as well as the sampling methodology applied for the monitoring program. The sam-

2001

2000

1999

1998

Wetland 2 Inlet 1997

Year

1 2 3 1 2 3 1 2 3 1 2 3 1 2 3

Event

778 37 57.3 632 215 211 478 148 547 271

5 20 31 353

2.1 3.6 2.9 5.5

TSS

1 6.7 4.2 2.7 12.1 15.5 2.1 4.4 2.2 4.2

BOD5

0.51 0.96 1.06 1.73

0.45 0.88 0.71 2.24 1.49 2.92 1.04 0.73 0.67 1.22

TKN

0.045 0.136 0.183 0.23

0.48 0.2 0.19 0.79 0.45 0.5 0.53 0.275 0.71 0.27

Total P

11,000

800 410

20,000 2000 10,000 40,000 200,000 30,000 6000 10,000 10,000 10,000

Fec. Strep. /100 mL

20 980 >20,000 4700

1790 613 4500 400,000 3000 28,000 28,000 31,000 3400 6200

E. coli /100 mL

0.0019 0.0057 0.0077 0.0168

0.0214 0.0045 0.0104 0.0358 0.0211 0.0311 0.0249 0.0123 0.228 0.0149

Cu

0.28 1.14 2.83 12

11.1 1.7 1.02 19.7 4.38 6.51 15.5 8 17.3 8.26

Fe

Table 3. Summary of contaminant concentrations at inlets to Wetland 2 and Wetland 3 (mg/L unless otherwise noted)

0.0032 0.0015 0.0044 0.013

0.0603 0.0102 0.0034 0.0468 0.0226 0.0358 0.016 0.0075 0.0572 0.0216

Pb

Continued

0.023 0.078 0.099 0.114

0.204 0.041 0.02 0.231 0.243 0.293 0.122 0.089 0.311 0.131

Zn


289

2001

2000

1999

1998

Wetland 3 Inlet 1997

Year

1 2 3 1 2 3 1 2 3 1 2 3 1 2 3

Event

Table 3. Continued

1.8 4.8 3.5 4.5 13 12.6 0.6 3.7 9.4 7.4 16 0.9 2.9 3 3.1

BOD5

2560 260 219 1310 508 175 1820 753 221 1110 1400 36 38 22 64

TSS

0.64 1.75 0.58 3.98 1.84 1.89 0.96 0.71 1.22 0.92 3.86 0.61 0.58 1.08 1.1

TKN

0.46 0.48 0.41 1.75 1 0.42 1.48 0.75 0.48 0.61 1.45 0.117 0.144 0.18 0.222

Total P

2300

60,000 3000 10,000 20,000 20,000 10,000 5000 5000 10,000 10,000 20,000 620 1200

Fec. Strep. /100 mL

5170 2410 20,000 300,000 600,000 60,000 20,000 8200 30,000 5400 36,000 280 2300 13,000 18,000

E. coli /100 mL

0.0346 0.0057 0.0101 0.0755 0.0323 0.014 0.0662 0.0339 0.164 0.0525 0.0648 0.0027 0.0081 0.0078 0.0108

Cu

17.8 1.97 1.21 42.6 13.7 4.83 53.2 22.6 7.76 35.5 42.1 1.69 2.21 1.36 3.37

Fe

0.0649 0.0066 0.008 0.0723 0.0171 0.0085 0.0372 0.0186 0.0116 0.0336 0.0357 0.0039 0.0035 0.0022 0.0056

Pb

0.232 0.041 0.045 0.443 0.162 0.08 0.236 0.232 0.101 0.221 0.227 0.009 0.0045 0.061 0.064

Zn

290 FARRELL AND SCHECKENBERGER

301 1.17 260 1.08 137–661 0.85–1.62

1.03 1.00 0.78–1.37

TKN

Wetland 3 Inlet Mean 4.1 Median 3.7 95% C.I. 2.5–6.5

TSS

133 213 60–294

BOD5

Wetland 2 Inlet Mean 3.8 Median 3.9 95% C.I. 2.6–5.6

Mean/ median

0.49 0.48 0.32–0.75

0.29 0.27 0.19–0.42

Total P

6952 10,000 3654–13,226

8862 10,000 3656–21,480

Fec. Strep. /100 mL Cu

14,879 0.0222 18,000 0.0323 5652–39,171 0.012–0.040

4464 0.0152 4500 0.0158 1231–16,186 0.008–0.027

E. coli /100 mL

0.0128 0.0144 0.0069–0.0236

Pb

7.89 0.0131 7.76 0.0116 3.9–16.0 0.0075–0.0228

4.63 7.22 2.4–9.0

Fe

Table 4. Comparison of contaminant concentrations to Wetland 2 and Wetland 3 (mg/L unless otherwise noted)

0.086 0.101 0.0451–0.1647

0.107 0.118 0.067–0.169

Zn


291

292


Fig. 3. Comparison of mean concentrations at inlets to Wetland 2 and Wetland 3 using lognormal statistical distribution.

Fig. 4. Comparison of median concentrations at inlets to Wetland 2 and Wetland 3 using lognormal statistical distribution.


293

pling methodology typically applied in the monitoring programs discussed in the literature either used automated samplers, which retrieved a volume of water for certain measured flows, or used rain barrels to collect the runoff from bridge structures. Both of the methodologies used in other programs clearly vary from the sampling methodology applied in the Dartnall Road Interchange monitoring program. Table 5 compares the observed values with those provided in the literature for highway runoff. The data in Table 5 indicate a high degree of variability in the loading data provided in the literature. The amount and type of pollutant found in highway runoff is affected by many factors, including traffic volume, type of traffic, local land use, and weather patterns (Barrett et al. 1993). The lower boundary of the concentrations of heavy metals and BOD5 at Wetland 2 and Wetland 3 are generally lower than the values provided in the literature, however remain within the same order of magnitude. The lowest values were obtained for the September storm event of 2000, which was at the end of a summer with above average rainfall which would have reduced the accumulated mass of contaminants on the highway surfaces. In general, the concentrations observed in the monitoring program are comparable to the concentrations provided in the literature for runoff from the same land use, thus validating the methodology and results observed at the inlets to the wetlands. Despite the different land use of the areas draining toward each facility, the comparable concentrations of TKN are consistent with literature values which indicate a negligible difference in event mean concentrations (EMC) of TKN for residential land use and highways (Dedman et al. 1998). However, highways generally contribute higher EMC for BOD5 and heavy metals (such as Cu, Fe, Pb, and Zn) than residential land use. The comparable values for BOD5, Fe, Pb, and Zn are likely a result of higher traffic density along the Lincoln Alexander Parkway west of the interchange—a portion of which drains to Wetland 3—compared to the traffic density within the interchange ramps which drain to Wetland 2. The increased concentrations due to traffic density could offset the lower concentrations from the residential area draining to Wetland 3. The results in Table 4 and the data in Table 3 indicate that fecal streptococcus concentrations at the inlet to Wetland 2 were more variable than the concentrations observed at the inlet to Wetland 3, which resulted in the higher mean concentration at Wetland 2. In general, however, there was minimal difference in the “long-term” fecal streptococcus concentrations at Wetland 2 and Wetland 3. This could be the result of ubiquitous strains which cannot be distinguished from true fecal strains by typical analytical procedures (Metcalf and Eddy Inc. 1991). By contrast, the concentration of E. coli at the inlet of Wetland 3 was generally greater than the concentration to Wetland 2. Given that E. coli is a product of animal waste, the higher concentrations observed at the inlet to Wetland 3 are likely due to the residential runoff entering the facility.

TSS BOD5 TKN Total P Cu Fe Pb Zn

Contaminant

5–778 1.0–15.5 0.45–2.92 0.045–0.79 0.0019–0.228 0.28–19.7 0.0015–0.0603 0.02–0.311

Wetland 2

22–2560 0.6–16 0.58–3.98 0.117–1.75 0.0027–0.164 1.21–53.2 0.0022–0.0723 0.0045–0.443

Wetland 3 45–798 12.7–37 0.335–55.0 0.113–0.998 0.022–7.033 2.429–10.3 0.073–1.78 0.056–9.29

Barrett et al. (1993)