Excerpted from: Pitt, R. Stormwater Quality Management, Part One: Drainage Design Philosophy, Effects, and Sources of Stormwater. CRC/Lewis. New York. Expected publication in 2000.

The main purpose of treating stormwater is to reduce its adverse impacts on receiving water beneficial uses. Therefore, it is important in any urban stormwater runoff study to assess the detrimental effects that runoff is actually having on a receiving water. Urban receiving waters may have many beneficial use goals, including:

· stormwater conveyance (flood prevention)

· biological uses (warm water fishery, biological integrity, etc.)

· non-contact recreation (linear parks, aesthetics, boating, etc.)

· contact recreation (swimming)

· water supply

With full development in an urban watershed and with no stormwater controls, it is unlikely that any of these uses can be obtained. With less development and with the application of stormwater controls, some uses may be possible. It is important that unreasonable expectations not be placed on urban waters, as the cost to obtain these uses may be prohibitive. With full-scale development and lack of adequate stormwater controls, severely degraded streams will be common. However, stormwater conveyance and aesthetics should be the basic beneficial use goals for all urban waters. Biological integrity should also be a goal, but with the realization that the natural stream ecosystem will be severely modified with urbanization. Certain basic controls, installed at the time of development, plus protection of stream habitat, may enable partial use of some of these basic goals in urbanized watersheds. Careful planning and optimal utilization of stormwater controls are necessary to obtain these basic goals in most watersheds. Water contact recreation, consumptive fisheries, and water supplies are not appropriate goals for most urbanized watersheds. However, these higher uses may be possible in urban areas where the receiving waters are large and drain mostly undeveloped areas.

Water Environment & Technology (1996a) reported that the latest National Water Quality Inventory released by the EPA only showed a slight improvement in the attainment of beneficial uses in the nations waters. Urban runoff was cited as the leading source of problems in estuaries, with nutrients and bacteria as the leading problems. Problems in rivers and lakes were mostly caused by agricultural runoff, with urban runoff the third ranked source for lakes, and the fourth ranked source for rivers. Bacteria, siltation, and nutrients were the leading problems in the nations rivers and lakes. Borchardt and Sperling (1996) stressed that many conditions may affect receiving waters from stormwater, specifically physical factors (such as shear stress) and chemical factors (such as oxygen depletion and/or non-ionized ammonia).

Integrated stormwater models that consider the collection system, treatment plant, and receiving water are rare. Rauch and Harremoës (1996) and Harremoës, et al. (1996) describe their efforts in model integrating, especially in evaluating receiving water wet weather dissolved oxygen and ammonia toxicity levels. Brelot, et al. (1996) also presented a decision analysis methodology for evaluating receiving water impacts from wet weather flows.

Pereira, et al. (1996) assessed the effects of human activities and land use on the water quality of the San Joaquin river and its major tributaries. This study focused on pesticides and organic contaminants and examined water, particulate forms, sediment, and a bivalve.

If just conventional water quality measures are used, almost all (87%) of Delaware’s non-tidal streams supported their designated biological uses. However, when biological assessments are included, only 13% of the streams were satisfactory, according to a review article by Claytor (1996c).

Weed invasion in the bushland surrounding the Lane Cove catchment in Australia was related to contamination of the floodplain by heavy metals and nutrients, plus increased flooding from the urban development surrounding the bushland, according to Riley and Banks (1996).

Stormwater infiltration into sanitary sewerage can have detrimental effects on sewage treatment facilities, as reported by Suresh, et al. (1996). The dilution of the sanitary wastewater adversely affects many unit operations.

In general, monitoring of urban stormwater runoff has indicated that the biological beneficial uses of urban receiving waters are most likely affected by habitat destruction and long-term pollutant exposures (especially to macroinvertebrates via contaminated sediment), while documented effects associated from acute exposures of toxicants in the water column are rare (Field and Pitt 1990; Pitt 1994; Pitt 1995). Receiving water pollutant concentrations resulting from runoff events and typical laboratory bioassay test results have not indicated many significant short-term receiving water problems. As an example, Lee and Jones-Lee (1996) state that exceedences of numeric criteria by short-term discharges do not necessarily imply that a beneficial use impairment exists. Many toxicologists and water quality expects have concluded that the relatively short periods of exposures to the toxicant concentrations in stormwater are not sufficient to produce the receiving water effects that are evident in urban receiving waters, especially considering the relatively large portion of the toxicants that are associated with particulates (Lee and Jones-Lee 1995). Lee and Jones-Lee (1995) conclude that the biological problems evident in urban receiving waters are mostly associated with illegal discharges and that the sediment bound toxicants are of little risk. Mancini and Plummer (1986) have long been advocates of numeric water quality standards for stormwater that reflect the partitioning of the toxicants and the short periods of exposure during rains. Unfortunately, this approach attempts to isolate individual runoff events and does not consider the accumulative adverse effects caused by the frequent exposures of receiving water organisms to stormwater (Davies 1995; Herricks, et al. 1996a and 1996b). Recent investigations have identified acute toxicity problems associated with moderate-term (about 10 to 20 day) exposures to adverse toxicant concentrations in urban receiving streams (Crunkilton, et al.

1996). However, the most severe receiving water problems are likely associated with chronic exposures to contaminated sediment and to habitat destruction.

Pathogens in stormwater are also a significant concern potentially affecting human health. The use of indicator bacteria is controversial for stormwater, as well as the assumed time of typical exposure of swimmers to contaminated receiving waters. However, recent epidemiological studies has shown significant health effects associated with stormwater contaminated marine swimming areas. Protozoa pathogens, especially associated with likely sewage-contaminated stormwater, is also of public health concern.

This chapter contains a summary of recent work describing the ecological and potential human health effects of stormwater. Also included is a brief review of the fates of stormwater toxicants and pathogens in receiving waters. It starts with a review of recent work describing recommended indicators that can be effectively used to assess receiving water problems in a local area. Chapter 4 describes sampling requirements and methods to characterize potential problem constituents in stormwater. However, biological monitoring methods are not described there, but are covered in detail in the related book by Burton and Pitt (1998). This chapter ends with a case study describing a comprehensive investigation of receiving water problems caused by stormwater.

Urban runoff has been found to cause significant receiving water impacts on aquatic life. The effects are obviously most severe for receiving waters draining heavily urbanized watersheds. However, some studies have shown important aquatic life impacts for streams in watersheds that are less than ten percent urbanized.

In order to best identify and understand these impacts, it is necessary to include biological monitoring, using a variety of techniques, and sediment quality analyses, in a monitoring program. Water column testing alone has been shown to be very misleading. Most aquatic life impacts associated with urbanization are probably related to long-term problems caused by polluted sediments and food web disruption. Transient water column quality conditions associated with urban runoff probably rarely cause significant aquatic life impacts.

The underlying theme of these researchers is that an adequate analysis of receiving water biological impacts must include investigations of a number of biological organism groups (fish, benthic macroinvertebrates, algae, rooted macrophytes, etc.) in addition to studies of water and sediment quality. Simple studies of water quality alone, even with possible comparisons with water quality criteria for the protection of aquatic life, are usually inadequate to predict biological impacts associated with urban runoff.

Duda, et al. (1982) presented a discussion on why traditional approaches for assessing water quality, and selecting control options, in urban areas have failed. The main difficulties of traditional approaches when used with urban runoff are: the complexity of pollutant sources, wet weather monitoring problems, and limitations when using water quality standards to evaluate the severity of wet weather receiving water problems. They also discuss the difficulty of meeting water quality goals in urban areas that were promulgated in the Water Pollution Control Act.

Relationships between observed receiving water biological effects and possible causes have been especially difficult to identify, let alone quantify. The studies reported in this paper have identified a wide variety of possible causative agents, including sediment contamination, poor water quality (low dissolved oxygen, high toxicants, etc.), and factors effecting the physical habitat of the stream (high flows, unstable streambeds, absence of refuge areas, etc.). It is expected that all of these factors are problems, but their relative importance varies greatly depending on the watershed and receiving water conditions. Horner (1991), as an example, notes that many watershed, site, and organism specific factors must be determined before the best combination of runoff control practices to protect aquatic life can be determined.

Heaney, et al. (1980) conducted a comprehensive evaluation of the existing literature pertaining to urban runoff effects on receiving waters. They found that well documented cases of receiving water detrimental effects were scarce. Through their review of many reports, they found several reasons to question the implied cause and effect relationships between urban runoff and receiving water conditions. Impacts that were attributed to urban runoff were probably caused, in many cases, by other water pollution sources (such as combined sewer overflows, agricultural nonpoint sources, etc.). One of the major difficulties encountered in their study was the definition of "problem" that had been used in the reviewed projects. They found that very little substantive data had been collected to document beneficial use impairments. In addition, urban runoff impacts are most likely to be associated with small receiving waters, while most of the existing urban water quality monitoring information exists for larger bodies of water. It was also very difficult for many researchers to isolate urban runoff effects from other water pollutant sources, such as municipal and industrial wastes. This was especially important in areas that had combined sewers that overflowed during wet weather contributing to the receiving water impacts during wet weather conditions.

Mulliss, et al. (1996) found that several wet weather discharge parameters regularly pose a serious threat with regard to freshwater aquatic life. Widera and Podraza (1996) investigated in-stream biological conditions and water quality during 52 CSO events in three years in a small stream near Essen, Germany. Notable observations were that ammonium concentrations increased by up to 70 times during CSO discharges, protozoa counts were significantly higher downstream of the CSO, while macroinvertebrate counts showed little difference. However, the composition of the aquatic life communities differed substantially between upstream and downstream locations, showing that common ecological indices (such as the index of diversity) are not suitable tools for detecting these changes because they do not correctly reflect the differences in community structure.

The time scale of biological impacts in receiving waters affected by stormwater must also be considered. Snodgrass, et al. (1998) reported that ecological responses to watershed changes may take between 5 and 10 years to equilibrate. Therefore, receiving water investigations conducted soon after disturbances or mitigation may not accurately reflect the long-term conditions that will eventually occur. They found that the first changes due to urbanization will be to stream and groundwater hydrology, followed by fluvial morphology, then water quality, and finally the aquatic ecosystem. They also reported that it is not possible to predict biological responses from in-stream habitat changes or conditions, although many researchers have concluded that habitat changes are the most serious cause of the aquatic biological problems associated with urbanization of a watershed.

Keefer, et al. (1979) examined the data from 104 water quality monitoring sites near urban areas throughout the country for dissolved oxygen conditions. These stations were selected from more than 1,000 nationwide monitoring stations operated by various federal and state agencies. They conducted analyses of daily dissolved oxygen data for 83 of these sites. About one half of the monitoring stations examined showed a 60 percent or greater, probability of a higher than average dissolved oxygen deficit occurring at times of higher than average streamflow, or on days with rainfall. This result was based upon daily data for entire water years; not all years at any given location exhibited this 60 percent probability condition. They found that the DO levels fell to less than 75 percent saturation at most of the stations that had this 60 percent or greater probability condition. They also found that DO concentrations of less than 5 mg/L were common. Keefer, et al. (1979) examined hourly dissolved oxygen data at 22 nationwide sites to find correlations between flows and DO deficit. They found that for periods of steady low flows, the DO fluctuated widely on a daily cycle, ranging from 1 to 7 mg/L. During rain periods, however, the flow increased, of course, but the diurnal cycle of this dissolved oxygen fluctuation disappeared. The minimum DO dropped from 1 to 1.5 mg/L below the minimum values observed during steady flows, and remained constant for periods ranging from 1 to 5 days. They also reported that as the high flow conditions ended, the DO levels resumed diurnal cyclic behavior. About 50 percent of the stations examined in detail on an hour by hour basis would not meet a 5 mg/L DO standard, and about 25 percent of these stations would not even meet a 2.0 mg/L standard for 4-hour averages. The frequency of these violations was estimated to be up to 5 times a year per station.

Another study that examined dissolved oxygen depletion on a regional basis was conducted by Ketchum (1978). This study examined dissolved oxygen data in Indiana. Sampling was conducted at nine cities and the project was designed to detect significant dissolved oxygen deficits in streams during periods of rainfall and runoff. The results of this study indicated that wet weather DO levels generally appear to be similar or higher than those observed during dry weather conditions in the same streams. They found that significant wet weather DO depletions were not observed, and due to the screening nature of the sampling program, more subtle impacts could not be measured. As noted below, adverse dissolved oxygen conditions associated with urban runoff are likely to occur a substantial time after the runoff event and downstream from the discharge locations.

Figure XX illustrates a problem that may be common to DO predictions in urban receiving waters. Pitt (1979) conducted three long-term BOD experiments with stormwater collected from a residential area in San Jose, CA. These were conventional BOD tests, using approved procedures published in the then current version of Standard Methods. Basically, many BOD bottles were prepared for each sample, representing replicates for each day for the observations, and for several different dilutions. The bottles were seeded with an activated sludge seed to provide a starting microbial population. As seen, the observed BOD curves do not have a conventional shape. The BOD₅ values are about 25 mg/L, typical to what is commonly reported for most stormwater. However, the BOD curves are seen to rapidly increase throughout the 20-day test period, instead of leveling off at about 7 to 10 days, as expected for municipal wastewaters. These curves illustrate the common problem of acclimation of a wastewater to the microörganisms that are present in the test solution. Stormwater has relatively low levels of nutrients and easily assimilated organic material, but moderate levels of toxicants. It is possible that the activated sludge seed requires extra time for the microbial population to shift to a population dominated by organisms capable of effectively degrading the organics in stormwater. Alternatively (or in addition), the more refractory organics in stormwater may simply require a longer period of time for degradation. In any case, the ultimate BOD/BOD₅ ratio for stormwater is much greater than for conventional municipal wastewaters, making simple use of observed BOD₅ values in receiving water models problematic. Urban stream sediments are commonly anaerobic, likely caused by the deposition of the slowly decaying stormwater organic compounds. Therefore, rapid stormwater effects on stream DO levels may be minimal, but sediment interaction (including scour) with the water can have adverse effects long after the stormwater event that discharged the decaying material.

Figure XXXX. Long-term BOD tests for stormwater (Pitt 1979)

Heaney, et al. (1980), during their review of studies that examined continuous dissolved oxygen (DO) monitoring stations downstream from urbanized areas, indicated that the worst dissolved oxygen levels occurred after the storms in about one-third of the cases studied. This lowered DO could be due to urban runoff moving downstream, combined sewer overflows and/or resuspension of benthic deposits. Resuspended benthic deposits could have been previously settled urban runoff settleable solids. They also found that worst case conditions do not always occur during the low flow periods following storms.

The main impact of a CSO is generally a decreased level of oxygen in the receiving waters, according to Seidl, et al. (1996). They conducted an extensive monitoring program in a Paris suburb to measure the bacteria and organic carbon content of a combined sewer under both wet and dry weather conditions in order to more accurately predict the resulting DO conditions. Lammersen (1996) examined dissolved oxygen and ammonia conditions in receiving waters affected by stormwater in northern Germany. She did not find any events during her three year monitoring period that may have caused critical conditions for these two parameters.

Numerous data are available characterizing stormwater chemical characteristics, and are summarized in Chapter 6. This discussion summarizes a few example cases where in-stream measurements found significant changes in quality as a function of land use. These studies usually sampled streams as they passed through urban areas, from upstream relatively uncontaminated areas through and past urban areas. Both wet and dry weather sampling was also usually conducted.

In the southeast, many urban lakes in developing areas are typically characterized by high turbidity levels caused by high erosion rates of fine grained clays. There has been conflicting evidence on the role of these elevated turbidity levels on eutrophication processes and resulting highly fluctuating DO levels. Because of the high sediment loads, these urban lakes are quite different compared to most studied impoundments. Burkholder, et al. (1998) described a series of enclosure experiments they conducted in Durant Reservoir, near Raleigh, North Carolina. The lake surface is about 5 ha, averages 2.5 m deep, and has a moderately urbanized watershed 350 ha in size. The hydraulic residence time ranges from as short as 7 days in the winter to 60 days in the summer. Secchi disk transparency ranged from 0.5 to 1.3 m during the summer of 1990 when these tests were conducted. The algal communities are P-limited until late summer, when N becomes the primary limiting nutrient. The phytoplankton biomass significantly increases during the summer growing season, with some nuisance algae (such as Anabaena spp.) occurring. Thermal stratification was obvious during the summer in the deeper areas of the lake, with the lower 0.5 m of water usually having low diurnal DO concentrations of £ 3.0 mg/L. Several 2 m diameter enclosures were constructed isolating sediment to water surface columns of water. The experimental design allowed investigating the effects of different levels of sediment and nutrients on algal productivity. They found that the effects (reduction of light reduction and coflocculation of clay and phosphate) of low (about 5 mg/L) and moderately high clay (about 15 mg/L) loadings added every 7 to 14 days did not significantly reduce the algal productivity simulation caused by high phosphate loadings. However, they noted that other investigators using higher clay loadings (about 25 mg/L added every 2 days) did see depressed effects of phosphorus enrichment on the test lake. They concluded that dynamically turbid systems, such as represented in southeastern urban lakes, have complex interacting mechanisms between discharged clay and nutrients that make simple predictions of the effects of eutrophication much more difficult than in the more commonly studied clear lakes. In general, increased turbidity will either have no effect, or will have a mitigating effect, on the cultural eutrophication process.

Field and Cibik (1980) summarized some potential urban runoff effects reported in other studies. Two studies of a reservoir near Knoxville, Tennessee, showed that the quality of the contributing streams were degraded to a small extent by urban runoff and that the reservoir itself experienced a significant change in DO, pH, BOD₅, conductivity, temperature, total solids, and total coliform bacteria during short storm events. In another study at the Christina River in Newark, Delaware, cadmium and lead concentrations several miles below the urban area remained at elevated values up to 48 hours after storm periods. The quality of runoff from similar non-urbanized watersheds was compared with this urbanized area's runoff. They found that concentrations of nitrates, phosphorus, heavy metals and pesticides were considerably higher in the urbanized areas than in the forested regions. Field and Cibik also reported on a study conducted in Virginia, where water, sediment, detritus, caddisflies, snails and crayfish were analyzed for iron, manganese, nickel, lead, cadmium, zinc, chromium and copper. The sampling areas were exposed to wastewater effluent and urban runoff. The found that concentrations increased immediately below stormwater discharge locations. They also reported on another study recently completed in Hawaii which indicated that receiving water conditions were hazardous because of suspended solids, heavy metals and bacterial pathogens.

During the Coyote Creek, San Jose study, dry weather concentrations of many constituents exceeded expected wet weather concentrations by factors of two to five times (Pitt and Bozeman 1982). These data are summarized in the case study at the end of this chapter. During dry weather, many of the major constituents (e.g., major ions, total solids, etc.) were significantly greater in both the urban and nonurban reaches. These constituents were all found in substantially lower concentrations in the urban runoff and in the rain. The rain and the resultant runoff apparently diluted the concentrations of these constituents in the creek during wet weather. Within the urban area, many constituents were found in greater concentrations during wet weather than during dry weather (Chemical Oxygen Demand, organic nitrogen, and especially heavy metals - lead, zinc, copper, cadmium, mercury, iron, and nickel). Lead concentrations were found to be more than seven times as great in the urban reach than in the nonurban reach during dry weather, with a confidence level of 75 percent. Other significant increases in urban area concentrations occurred for nitrogen, chloride, orthophosphate, Chemical Oxygen Demand, specific conductance, sulfate, and zinc. The dissolved oxygen measurements were about 20 percent less in the urban reach than in the nonurban reach of the creek.

Bolstad and Swank (1997) examined the in-stream water quality at 5 sampling stations in Cowetta Creek in western North Carolina over a 3 year period. The watershed is 4350 ha and is relatively undeveloped (forested) in the area above the most upstream sampling station and becomes more urbanized at the downstream sampling station. Baseflow water quality was good, while most constituents increased during wet weather. Bacteria values increased substantially during wet weather, with total and fecal coliforms, and fecal streptococci increasing by two to three times during storms. Water quality was compared to building density for the different monitoring stations, with increasing stormwater pollutant concentrations (especially for turbidity, bacteria, and some inorganic solutes) with increasing building densities. Baseflow concentrations also typically increased with density, but at a much lower rate. In addition, the highest concentrations observed during individual events corresponded to the highest flow rates.

Reported Fish Kill Information

Urban runoff impacts are sometimes difficult for many people to appreciate in urban areas. Fish kills are the most obvious indication of water quality problems for many people. However, because urban receiving water quality is usually so poor, the aquatic life in typical urban receiving waters is usually limited in abundance and diversity, and quite resistant to poor water quality. Sensitive native organisms have typically been displaced, or killed, long ago. It is also quite difficult to identify the specific cause of a fish kill in an urban stream. Ray and White (1979), for example, stated that one of the complicating factors in determining fish kills related to heavy metals is that the fish mortality may lag behind the first toxic exposure by several days, and is usually detected many miles downstream from the discharge location. The actual concentrations of the water quality constituents that may have caused the kill could then be diluted beyond detection limits, making probable sources of the toxic materials impossible to determine in many cases.

Heaney, et al. (1980) reviewed fish kill information reported to government agencies during 1970 to 1979. They found that less than three percent of the reported 10,000 fish kills were identified as having been caused by urban runoff. This is less than 30 fish kills per year nationwide. A substantial number of these 10,000 fish kills were not identified as having any direct cause. It is expected that many of these fish kills could have been caused by urban runoff, or a combination of problems that could have been worsened by urban runoff.

During the Bellevue, Washington, receiving water studies, some fish kills were noted in the unusually clean urban streams (Pitt and Bissonnette 1984). The fish kills were usually associated with inappropriate discharges to the storm drainage system (such as cleaning materials and industrial chemical spills) and not from "typical" urban runoff. However, as noted later, the composition of the fish in the urban stream was quite different, as compared to the control stream (Scott, et al. 1986).

Fish kill data have therefore not been found to be a good indication of receiving water problems caused by urban runoff. However, as discussed previously, the composition of the fisheries and other aquatic life taxonomic indicators are sensitive indicators of receiving water problems in urban streams.

Even though acute toxicity of stormwater on most aquatic organisms has been relatively rare, short-term toxicity tests are still commonly conducted as part of some whole effluent toxicity (WET) tests required by some state regulatory agencies and by some stormwater researchers.

The need for endpoints for toxicological assessments using multiple stressors was discussed by Marcy and Gerritsen (1996). They used five watershed-level ecological risk assessments to develop appropriate endpoints based on specific project objectives. Dyer and White (1996) also examined the problem of multiple stressors affecting toxicity assessments. They felt that field surveys rarely can be used to verify simple single parameter laboratory experiments. They developed a watershed approach integrating numerous databases in conjunction with in-situ biological observations to help examine the effects of many possible causative factors. Toxic effect endpoints are additive for compounds having the same "mode of toxic action", enabling predictions of complex chemical mixtures in water, as reported by Environmental Science & Technology (1996a). According to EPA researchers at the Environmental Research Laboratory in Duluth, MN, there are about five or six major action groups that contain almost all of the compounds of interest in the aquatic environment. Much work still needs to be done, but these new developing tools may enable the in-stream toxic effects of stormwater to be better predicted.

Ireland, et al. (1996) found that exposure to UV radiation (natural sunlight) increased the toxicity of PAH contaminated urban sediments to C. dubia. The toxicity was removed when the UV wavelengths did not penetrate the water column to the exposed organisms. Toxicity was also reduced significantly in the presence of UV when the organic fraction of the stormwater was removed. Photo-induced toxicity occurred frequently during low flow conditions and wet weather, but was reduced during turbid conditions.

Johnson, et al. (1996) and Herricks, et al. (1996a and 1996b) describe a structured tier testing protocol to assess both short-term and long-term wet weather discharge toxicity that they developed and tested. The protocol recognizes that the test systems must be appropriate to the time-scale of exposure during the discharge. Therefore, three time-scale protocols were developed, for intra-event, event, and long-term exposures. The use of standard whole effluent toxicity (WET) tests were found to over-estimate the potential toxicity of stormwater discharges.

The effects of stormwater on Lincoln Creek, near Milwaukee, WI, were described by Crunkilton, et al. (1996). Lincoln Creek drains a heavily urbanized watershed of 19 mi² that is about nine miles long. On-site toxicity testing was conducted with side-stream flow-through aquaria using fathead minnows, plus in-stream biological assessments, along with water and sediment chemical measurements. In the basic tests, Lincoln Creek water was continuously pumped through the test tanks, reflecting the natural changes in water quality during both dry and wet weather conditions. The continuous flow-through mortality tests indicated no toxicity until after about 14 days of exposure, with more than 80% mortality after about 25 days, indicating that short-term toxicity tests likely underestimate stormwater toxicity. The biological and physical habitat assessments also supported a definitive relationship between degraded stream ecology and urban runoff.

Rainbow (1996) presented a detailed overview of heavy metals in aquatic invertebrates. He concluded that the presence of a metal in an organism cannot tell us directly whether that metal is poisoning the organism. However, if compared to concentrations in a suite of well-researched biomonitors, it is possible to determine if the accumulated concentrations are atypically high, with a possibility that toxic effects may be present. Allen (1996) also presented an overview of metal contaminated aquatic sediments. This book presents many topics that would enable the user to better interpret measured heavy metal concentrations in urban stream sediments.

One of the key objectives of the Chesapeake Bay restoration effort is to reduce the impacts of toxicants. Alden and Hall (1996) and Hall, et al. (1996) describe the Toxics Reduction Strategy, based on water column and sediment chemical analyses, benthic community health, and fish body burdens. More than 40% of the sites have displayed some degree of water column toxicity, and about 70% of the sites have displayed sediment toxicity. Garries, et al. (1996) further describe how the list of Toxics of Concern is developed for Chesapeake Bay.

Stormwater impacts to streams are not limited to the relatively short duration of runoff events. As an example, sediments can dominate the aquatic physiochemical and biological processing of nutrients. In addition, sediment contaminated by stormwater pollutants has a detrimental effect on the receiving water biological community. Schueler (1996b) summarized in-situ assessment methods of stormwater-impacted sediments. The use of in-situ test chambers, using C. dubia, eliminates many of the sample disruption problems associated with conducting sediment toxicity tests in the laboratory.

Pitt (1996) evaluated various laboratory toxicity tests using 20 stormwater and CSO samples. He found that the most promising results are associated with using several complementary tests, instead of any one test method. However, simple screening toxicity tests (such as using the Azur Microtox test) are useful during preliminary assessments or for treatability tests.

Many studies have shown the severe detrimental effects of urban runoff on receiving water organisms. These studies have generally examined receiving water conditions above and below a city, or by comparing two parallel streams, one urbanized and another nonurbanized. The researchers usually carefully selected the urbanized streams to minimize pollutant sources other than urban runoff. However, few studies have examined direct cause and effect relationships of urban runoff for receiving water aquatic organisms (Heaney and Huber 1984). The following paragraphs briefly describe a variety of urban receiving water investigations.

Klein (1979) studied 27 small watersheds having similar physical characteristics, but having varying land uses, in the Piedmont region of Maryland. During an initial phase of the study, they found definite relationships between water quality and land use. Subsequent study phases examined aquatic life relationships in the watersheds. The principal finding was that stream aquatic life problems were first identified with watersheds having imperviousness areas comprising at least 12 percent of the watershed. Severe problems were noted after the imperviousness quantities reached 30 percent.

Receiving water impact studies were also conducted in North Carolina (Lenet, et al. 1979; Lenet and Eagleson 1981; Lenat, et al. 1981). The benthic fauna occurred mainly on rocks. As sedimentation increased, the amount of exposed rocks decreased, with a decreasing density of benthic macroinvertebrates. Data from 1978 and 1979 in five cities showed that urban streams were grossly polluted by a combination of toxicants and sediment. Chemical analyses, without biological analyses, would have underestimated the severity of the problems because the water column quality varied rapidly, while the major problems were associated with sediment quality and effects on macroinvertebrates. Macroinvertebrate diversities were severely reduced in the urban streams, compared to the control streams. The biotic indices indicated very poor conditions for all urban streams. Occasionally, high populations of pollutant tolerant organisms were found in the urban streams, but would abruptly disappear before subsequent sampling efforts. This was probably caused by intermittent discharges of spills or illegal dumpings of toxicants. Although the cities studied were located in different geographic areas of North Carolina, the results were remarkably uniform.

During the Coyote Creek, San Jose, California, receiving water study (described later in a case study discussion), 41 stations were sampled in both urban and nonurban perennial flow stretches of the creek over three years. Short and long-term sampling techniques were used to evaluate the effects of urban runoff on water quality, sediment properties, fish, macroinvertebrates, attached algae, and rooted aquatic vegetation (Pitt and Bozeman 1982). These investigations found distinct differences in the taxonomic composition and relative abundance of the aquatic biota present. The non-urban sections of the creek supported a comparatively diverse assemblage of aquatic organisms including an abundance of native fishes and numerous benthic macroinvertebrate taxa. In contrast, however, the urban portions of the creek (less than 5% urbanized), affected only by urban runoff discharges and not industrial or municipal discharges, had an aquatic community generally lacking in diversity and was dominated by pollution-tolerant organisms such as mosquitofish and tubificid worms.

A major nonpoint runoff receiving water impact research program was conducted in Georgia (Cook, et al. 1983). Several groups of researchers examined streams in major areas of the state. Benke, et al. (1981) studied 21 stream ecosystems near Atlanta having watersheds of one to three square miles each and land uses ranging from 0 to 98 percent urbanization. They measured stream water quality but found little relationship between water quality and degree of urbanization. The water quality parameters also did not identify a major degree of pollution. In contrast, there were major correlations between urbanization and the number of species found. They had problems applying diversity indices to their study because the individual organisms varied greatly in size (biomass). CTA (1983) also examined receiving water aquatic biota impacts associated with urban runoff sources in Georgia. They studied habitat composition, water quality, macroinvertebrates, periphyton, fish, and toxicant concentrations in the water, sediment, and fish. They found that the impacts of land use were the greatest in the urban basins. Beneficial uses were impaired or denied in all three urban basins studied. Fish were absent in two of the basins and severely restricted in the third. The native macroinvertebrates were replaced with pollution tolerant organisms. The periphyton in the urban streams were very different from those found in the control streams and were dominated by species known to create taste and odor problems.

Pratt, et al. (1981) used basket artificial substrates to compare benthic population trends along urban and nonurban areas of the Green River in Massachusetts. The benthic community became increasing disrupted as urbanization increased. The problems were not only associated with times of heavy rain, but seemed to be affected at all times. The stress was greatest during summer low flow periods and was probably localized near the stream bed. They concluded that the high degree of correspondence between the known sources of urban runoff and the observed effects on the benthic community was a forceful argument that urban runoff was the causal agent of the disruption observed.

Cedar swamps in the New Jersey Pine Barrens were studied by Ehrenfeld and Schneider (1983). They examined nineteen wetlands subjected to varying amounts of urbanization. Typical plant species were lost and replaced by weeds and exotic plants in urban runoff affected wetlands. Increased uptakes of phosphorus and lead in the plants were found. It was concluded that the presence of stormwater runoff to the cedar swamps caused marked changes in community structure, vegetation dynamics, and plant tissue element concentrations.

Medeiros and Coler (1982) and Medeiros, et al. (1984) used a combination of laboratory and field studies to investigate the effects of urban runoff on fathead minnows. Hatchability, survival, and growth were assessed in the laboratory in flow-through and static bioassay tests. Growth was reduced to one half of the control growth rates at 60 percent dilutions of urban runoff. The observed effects were believed to be associated with a combination of toxicants.

The University of Washington (Pederson 1981; Richey, et al. 1981; Perkins 1982; Richey 1982; Scott, et al. 1982; Ebbert, et al. 1983; Pitt and Bissonnette 1984; and Prych and Ebbert undated) conducted a series of studies to contrast the biological and chemical conditions in urban Kelsey Creek with rural Bear Creek in Bellevue, Washington. The urban creek was significantly degraded when compared to the rural creek, but still supported a productive, but limited and unhealthy salmonid fishery. Many of the fish in the urban creek, however, had respiratory anomalies. The urban creek was not grossly polluted, but flooding from urban developments had increased dramatically in recent years. These increased flows dramatically changed the urban stream's channel, by causing unstable conditions with increased stream bed movement, and by altering the availability of food for the aquatic organisms. The aquatic organisms were very dependent on the few relatively undisturbed reaches. Dissolved oxygen concentrations in the sediments depressed embryo salmon survival in the urban creek. Various organic and metallic priority pollutants were discharged to the urban creek, but most of them were apparently carried through the creek system by the high storm flows to Lake Washington. The urbanized Kelsey Creek also had higher water temperatures (probably due to reduced shading) than Bear Creek. This probably caused the faster fish growth in Kelsey Creek.

The fish population in the urbanized Kelsey Creek had adapted to its degrading environment by shifting the species composition from coho salmon to less sensitive cutthroat trout and by making extensive use of less disturbed refuge areas. Studies of damaged gills found that up to three-fourths of the fish in Kelsey Creek were affected with respiratory anomalies, while no cutthroat trout and only two of the coho salmon sampled in the forested Bear Creek had damaged gills. Massive fish kills in Kelsey Creek and its tributaries were also observed on several occasions during the project due to the dumping of toxic materials down the storm drains.

There were also significant differences in the numbers and types of benthic organisms found in urban and forested creeks during the Bellevue research. Mayflies, stoneflies, caddisflies, and beetles were rarely observed in the urban Kelsey Creek, but were quite abundant in the forested Bear Creek. These organisms are commonly regarded as sensitive indicators of environmental degradation. One example of degraded conditions in Kelsey Creek was shown by a specie of clams (Unionidae) that was not found in Kelsey Creek, but was commonly found in Bear Creek. These clams are very sensitive to heavy siltation and unstable sediments. Empty clam shells, however, were found buried in the Kelsey Creek sediments indicating their previous presence in the creek and their inability to adjust to the changing conditions. The benthic organism composition in Kelsey Creek varied radically with time and place while the organisms were much more stable in Bear Creek.

XXXX figures/tables from Bellevue XXXXXX

Urban runoff impact studies were conducted in the Hillsborough River near Tampa Bay, Florida, as part of the U.S. EPA’s Nationwide Urban Runoff Program (NURP) (Mote Marine Laboratory 1984). Plants, animals, sediment, and water quality were all studied in the field and supplemented by laboratory bioassay tests. Effects of salt water intrusion and urban runoff were both measured because of the estuarine environment. During wet weather, freshwater species were found closer to the Bay than during dry weather. In coastal areas, these additional natural factors made it even more difficult to identify the cause and effect relationships for aquatic life problems. During another NURP project, Striegl (1985) found that the effects of accumulated pollutants in Lake Ellyn (Glen Ellyn, Ill.) inhibited desirable benthic invertebrates and fish and increased undesirable phyotoplankton blooms.

The number of benthic organism taxa in Shabakunk Creek in Mercer County, New Jersey, declined from 13 in relatively undeveloped areas to four below heavily urbanized areas (Garie and McIntosh 1986 and 1990). Periphyton samples were also analyzed for heavy metals with significantly higher metal concentrations found below the heavily urbanized area than above.

A number of papers presented at the 7^th International Conference on Urban Storm Drainage, held in Hannover, Germany, described receiving water studies that investigated organic and heavy metal toxicants. Handová, et al. (1996) examined the bioavailability of metals from CSOs near Prague. They compared these results with biomonitoring. The metals were ranked according to their mobility as: Cd (95%), Zn (87%), Ni (64%), Cr (59%), Pb (48%), and Cu (45%). The mobile fraction was defined as the metal content that was exchangeable, bound to carbonates, bound to iron and manganese oxides, and bound to organic matter. Boudries, et al. (1996) and Estèbe, et al. (1996) investigated heavy metals and organics bound to particulates in the River Seine near Paris. The Paris CSOs caused a significant increase in the aliphatic and aromatic hydrocarbons bound to river sediments. The high flows during the winter were associated with lower heavy metal associations with the sediment, compared to the lower summer flow conditions. These differences were found to be due to dilution of the CSOs in the river and to the changing contributions of rural versus urban suspended solids during the different seasons.

The Northeastern Illinois Planning Commission (Dreher 1997) compared comprehensive fish survey information from over 40 northeastern Illinois small to moderate-sized streams and rivers to demographic data for the contributing watershed areas. The streams had watershed areas ranging from about 12 to 222 square miles and had population densities ranging from about 30 to more than 4,500 people per square mile. The fish data was used in the index of biotic integrity (IBI) to identify the quality of the fish populations. Table 1 lists the fish data that is used in the IBI and Table 2 shows the different scores for the quality categories. Factors necessary for good and excellent quality fish communities include the presence of diverse and reproducing fish and other aquatic organisms, including a significant percentage of intolerant species (such as darters and smallmouth bass).

The more commonly used imperviousness-based indicator of development was not used due to a lack of available data and the difficulty of acquiring good quality current imperviousness data, let alone estimating historical imperviousness data. In contrast, population data was readily available and thought to be an adequate indicator of the extent and density of urbanization in the watersheds. Figure 3 is a plot of the IBI vs. the population densities. They found that nearly all streams in urban and suburban watersheds having population densities greater than about 300 people per square mile showed signs of considerable impairment to their fish communities (being in fair to very poor condition). In contrast, nearly all rural streams supported fish communities that were rated good or excellent. They identified both point and nonpoint sources as major contributors to these impairments. However, the point source discharges and CSO discharges have substantially decreased over the past 20 years, while the nonpoint source discharges have increased significantly with increased development, and the fisheries are still declining in many areas. In stable areas that were mostly affected by point sources and CSOs, documented dramatic improvements in some water quality indicators (especially DO and ammonia), and the fish populations, have occurred. In similar areas, but having continued urban development, the fisheries have continued to decline.

Figure 3. Index of Biotic Integrity versus Population Density in Northeastern Illinois Streams (Dreher 1997).

They concluded that although rural watersheds have known water quality problems (especially agricultural chemicals and erosion, plus manure runoff), these issues did not prevent the attainment of mostly high quality fisheries in these areas. Similar conclusions were noted in the comparison study by the USGS in North Carolina (as reviewed by Schueler 1997b) of forested, agricultural, and urban streams. Although the forested streams were of the best quality, the streams in the agricultural areas were of intermediate quality and had significantly better biological conditions than the urban stream (which had poor macroinvertebrate and fish conditions, poor sediment and temperature conditions, and fair substrate and nutrient conditions).

Some of the most serious effects of urban runoff are on the aquatic habitat of the receiving waters. These habitat effects are in addition to the pollutant concentration effects. Numerous papers already referenced found significant sedimentation problems in urban receiving waters. The major effects of urban sediment on the aquatic habitat include silting of spawning and food production areas and unstable bed conditions (Cordone and Kelley 1961). Other major habitat destruction problems include rapidly changing flows and the absence of refuge areas to protect the biota during these flow changes. Removal of riparian vegetation can increase water temperatures and a major source of large organic debris that are important refuge areas. The major references on stream geomorphology that many of the following researchers based their work on were by Leopold, et al. (1964), Brookes (1988), and Rosgen (1994). These fundamental references should be consulted for excellent descriptions of the many natural processes affecting streams in transition. Brookes also specifically examines urbanization effects on stream morphology. Knowledge of these basic processes will better enable an understanding of local stream changes occurring with watershed urbanization. This understanding will, in turn, enable more efficient rehabilitation efforts of degraded streams and the use of watershed controls to minimize these effects. Stream rehabilitation efforts to improve stream habitat are discussed in Chapter 13.

Brookes (1988) has documented many cases in the U.S. and Great Britain of stream morphological changes associated with urbanization. These changes are mostly responsible for habitat destruction that are usually the most significant detriment to aquatic life. In many cases, water quality improvement would result in very little aquatic life benefits if the physical habitat is grossly modified. The most obvious habitat problems are associated with stream "improvement" projects, ranging from removal of debris, to straightening streams, to channelization projects. Brookes (1988 and 1991) presents a number of ways to minimize habitat problems associated with stream channel projects, including stream restoration.

Wolman and Schick (1967) observed deposition of channel bars, erosion of channel banks, obstruction of flows, increased flooding, shifting of channel bottoms, along with concurrent changes in the aquatic life, in Maryland streams affected by urban construction activities. Robinson (1976) studied eight streams in watersheds undergoing urbanization and found that the increased magnitudes and frequencies of flooding, along with the increased sediment yields, had considerable impact on stream morphology (and therefore aquatic life habitat).

Increased flows are the probably the best know example of impacts associated with urbanization. Most of the recognition has of course focused on increased flooding and associated damages. This has led to numerous attempts to control peak flows from new urban areas through the use of regulations that limit post development peak flows to pre development levels for relatively large design storms. The typical response has been to use dry detention ponds. Chapter 10 describes how this approach is limited, and may actually increase downstream flows. In addition to the serious issue of flooding, high flows also cause detrimental ecological problems in receiving waters. The following discussion presents several case studies where increased flows were found to have serious effects on stream habitat conditions.

The aquatic organism differences in urbanized and control streams found during the Bellevue Urban Runoff Program were probably most associated with the increased peak flows. The increased flows in the urbanized Kelsey Creek resulted in increases in sediment carrying capacity and channel instability of the creek (Pederson 1981; Perkins 1982; Richey, et al. 1981; Richey 1982; Scott, et al . 1982). Kelsey Creek had much lower flows than the control Bear Creek during periods between storms. About 30 percent less water was available in Kelsey Creek during the summers. These low flows may also have significantly affected the aquatic habitat and the ability of the urban creek to flush toxic spills or other dry weather pollutants from the creek system (Ebbert, et al. 1983; Prych and Ebbert undated). Kelsey Creek had extreme hydrologic responses to storm. Flooding substantially increased in Kelsey Creek during the period of urban development; the peak annual discharges almost doubled in the last 30 years, and the flooding frequency also increased due to urbanization (Ebbert, et al. 1983; Prych and Ebbert undated). These increased flows in urbanized Kelsey Creek resulted in greatly increased sediment transport and channel instability.

Bharuri, et al. (1997) also quantified the changes in streamflow and associated decreases in groundwater recharge associated with urbanization. They point out that the most widely addressed hydrologic effect of urbanization is the peak discharge increases that cause local flooding. However, the increase in surface runoff volume also represents a net loss in groundwater recharge. They point out that urbanization is linked to increased variability in volume of water available for wetlands and small streams, causing "flashy" or "flood-and-drought" conditions. In northern Ohio, urbanization at a study area was found to cause a 195% increase in the annual volume of runoff, while the expected increase in the peak flow for the local 100-yr event was only 26% for the same site. Although any increase in severe flooding is problematic and cause for concern, the much larger increase in annual runoff volume, and associated decrease in groundwater recharge, likely has a much greater effect on in-stream biological conditions.

Snodgrass, et al. (1998) reported that in the Toronto, Ontario, area, flows causing bankfull conditions occur with a return frequency of about 1.5 years. Storms with this frequency are in general equilibrium with resisting forces that tend to stabilize the channel (such as vegetation and tree root mats), with increased flows overcoming these resisting forces causing channel enlargement. Infrequent flows can therefore be highly erosive. With urbanization, the flows that were bankfull flows during historical times now occur much more frequently (about every 0.4 years in Toronto). The channel cross-sectional area therefore greatly increases to accommodate the increased stream discharges and power associated with the "new" 1.5 year flows that are trying to re-establish equilibrium.

Booth and Jackson (1997) examined numerous data from lowland streams in western Washington and concluded that development having about 10% imperviousness caused a readily apparent degradation of aquatic life in the receiving waters. They linked the association between increased imperviousness and biological degradation to increases in flows and sediment discharges. They concluded that conventional methods to size stormwater mitigation measures (especially detention ponds) were seriously inadequate. They felt that without a better understanding of the critical processes that lead to degradation, some downstream damage to the aquatic ecosystem is likely inevitable, without unpopular restrictions to the extent of development in the watershed corresponding to <10% imperviousness. Figure 4 is a plot showing channel stability as a function of effective imperviousness in a watershed, for different flow ratios between the historical 10-year predevelopment (forested) flow conditions and the current (urbanized) 2-year flow conditions. The stream channels were generally stable if the effective impervious areas remained below 10% of the complete watershed. This level of development corresponds to a 2-year developed condition flow being less than the historical 10-year pre-developed flow condition. They found that the classical goal of detention ponds to maintain predevelopment flows was seriously inadequate because there is no control on the duration of the peak flows. They showed that a duration standard to maintain post development flow durations for all sediment-transporting discharges to predevelopment durations will avoid many receiving water habitat problems associated with stream instability. Without infiltration, the amount of runoff will obviously still increase with urbanization, but the increased water could be discharged from detention facilities at flow rates below the critical threshold causing sediment transport. The identification of the threshold discharge below which sediment transport does not occur, unfortunately, if difficult and very site specific. A presumed threshold discharge of about one-half of the pre-development 2-year flow was recommended for gravel bedded streams. Sand-bedded channels have sediment transport thresholds that are very small, with inevitable bed load transport likely to occur for most levels of urbanization.

Figure 4. Channel Stability and Effective Imperviousness for Historical and Current Flow Ratios (Booth and Jackson 1997).

Changes in physical stream channel characteristics can have a significant effect on the biological health of the stream. Schueler (1996) stated that channel geometry stability can be a good indicator of the effectiveness of stormwater control practices. He also found that once a watershed area has more than about 10 to 15% effective impervious cover, noticeable changes in channel morphology occur, along with quantifiable impacts on water quality, and biological conditions. Stephenson (1996) studied changes in streamflow volumes in South Africa during urbanization. He found increased stormwater runoff, decreases in the groundwater table, and dramatically decreased times of concentration. The peak flow rates increased by about two-fold, about half caused by increased pavement (in an area having only about 5% effective impervious cover), with the remainder caused by decreased times of concentration.

Richey (1982) made some observations about bank stabilities in Kelsey and Bear Creeks as part of the Bellevue, WA, NURP project (Pitt and Bissonnette 1984). She notes that the Kelsey Creek channel width had been constrained during urban development. Thirty-five percent of the urbanized Kelsey Creek channel mapped during these projects was modified by the addition of some type of stabilization structure. Only eight percent of non-urbanized Bear Creek’s length was stabilized. Most of the stabilization structures in Bear Creek were low walls in disrepair while more than half of the structures observed along Kelsey Creek were large riprap or concrete retention walls. The necessity of the stabilization structures was evident from the extent and severity of erosion cuts and the number of deposition bars observed along the Kelsey Creek stream banks. Bridges and culverts were also frequently found along Kelsey Creek; these structures further act to constrict the channel. As discharges increased and the channel width is constrained, the velocity increases, causing increases in erosion and sediment transport.

The use of heavy riprapping along the creek seemed to worsen the flood problems. Storm flows are unable to spread out onto the flood plain and the increased velocities are evident downstream along with increased sediment loads. This rapidly moving water has enough energy to erode unprotected banks downstream of riprap. Many erosion cuts along Kelsey Creek downstream of these riprap structures were found. Similar erosion of the banks did not occur in Bear Creek. Much of the Bear Creek channel had a wide flood plain with many side sloughs and back eddies. High flows in Bear Creek could spread onto the flood plains and drop much of their sediment load as the water velocities decreased.

The University of Washington studies also examined sediment transport in urbanized Kelsey and non-urbanized Bear Creeks. Richey (1982) found that the relative lack of debris dams and off-channel storage areas and sloughs in Kelsey Creek contributed to the rapid downstream transit of water and materials. The small size of the riparian vegetation and the increased stream power probably both contributed to the lack of debris in the channel. It is also possible that the channel debris may have been cleared from the stream to facilitate rapid drainage. The high flows from high velocities caused the sediments to be relatively coarse. The finer materials were more easily transported downstream. Larger boulders were also found in the sediment but were probably from failed riprap or gabion structures.

Maxted (1997) examined stream problems in Delaware associated with urbanization. He found an apparent strong correlation between habitat score and biology score from 40 stream study locations (Figure 72.1). This figure shows that it is not possible to have acceptable biological conditions if the habitat is degraded. The leading contributor to habitat degradation was found to be urban runoff, especially the associated high flows and sediment accumulations.

Figure 72.1. Relationship between habitat and biological conditions in small Delaware streams (Claytor 1996).

A number of presentations concerning aquatic habitat effects from urbanization were made at the Effects of Watershed Development and Management on Aquatic Ecosystems conference held in Snowbird, UT, in August of 1996, sponsored by the Engineering Foundation and the ASCE. MacRae (1997) presented a review of the development of the common zero runoff increase (ZRI) discharge criterion, referring to peak discharges before and after development. This criterion is commonly met using detention ponds for the 2 yr storm. MacRae shows how this criterion has not effectively protected the receiving water habitat. He found that stream bed and bank erosion is controlled by the frequency and duration of the mid-depth flows (generally occurring more often than once a year), not the bank-full condition (approximated by the 2 yr event). During monitoring near Toronto, he found that the duration of the geomorphically significant pre-development mid-bankfull flows increased by a factor of 4.2 times, after 34% of the basin had been urbanized, compared to before development flow conditions. The channel had responded by increasing in cross-sectional area by as much as 3 times in some areas, and was still expanding. Table II shows the modeled durations of critical discharges for predevelopment conditions, compared to current and ultimate levels of development with "zero runoff increase" controls in place. At full development and even with full ZRI compliance in this watershed, the hours exceeding the critical mid-bankfull conditions will increase by a factor of 10, with resulting significant effects on channel stability and the physical habitat.

MacRae (1997) also reported other studies that found that channel cross-sectional areas began to enlarge after about 20 to 25% of the watershed was developed, corresponding to about a 5% impervious cover in the watershed. When the watersheds are completely developed, the channel enlargements were about 5 to 7 times the original cross-sectional areas. Changes from stable streambed conditions to unstable conditions appear to occur with basin imperviousness of about 10%, similar to the value reported previously for serious biological degradation. He also summarized a study conducted in British Columbia that examined 30 stream reaches in natural areas, in urbanized areas having peak flow attenuation ponds, and in urbanized areas not having any stormwater controls. The channel widths in the uncontrolled urban streams were about 1.7 times the widths of the natural streams. The streams having the ponds also showed widening, but at a reduced amount compared to the uncontrolled urban streams. He concluded that an effective criterion to protect stream stability (a major component of habitat protection) must address mid-bankfull events, especially by requiring similar durations and frequencies of stream power (the product of shear stress and flow velocity, not just flow velocity alone) at these depths, compared to satisfactory reference conditions.

Much research on habitat changes and rehabilitation attempts in urban streams has occurred in the Seattle area of western Washington over the past 20 years. Sovern and Washington (1997) described the in-stream processes associated with urbanization in this area, as part of a paper describing a recommended approach for the rehabilitation of urban streams. They were concerned that many "restoration" attempts of urban streams were destined to failure because of a lack of understanding of the actual changes occurring in streams as the watersheds changed from forested to urban land uses. They presented a concept of the "new urban stream" that attempts to correct several of the most important changes to better accommodate the native Pacific Northwest fish, instead of the unrealistic goal of trying to totally restore the steams to predevelopment conditions. The important factors that affect the direction and magnitude of the changes in a steam’s physical characteristics due to urbanization include:

In their natural state, small streams in forested watersheds in Western Washington have small low-flow channels (the aquatic habitat channel) with little meandering (Sovern and Washington 1997). The stream banks are nearly vertical because of clayey bank soils and heavy root structures, and the streams have numerous debris jams from fallen timber. The widths are also narrow, generally from 3 to 6 feet wide. Stable forested watersheds also support about 250 aquatic plant and animal species along the stream corridor. Pool/riffle habitat is dominant along streams having gradients less than about 2 percent slope, while pool/drop habitat is dominant along streams having gradients from 4 to 10 percent. The pools form behind large organic debris (LOD) or rocks. The salmon and trout in Western Washington have evolved to take advantage of these stream characteristics. Sovern and Washington (1997) point out that less athletic fish species (such as chum and pink salmon) cannot utilize the steeper gradient, upper reaches, of the streams. Coho, steelhead and cutthroat can use these upper areas, however.

Urbanization radically affects many of these natural stream characteristics. Pitt and Bissonnette (1984) reported that the coho and cutthroat were affected by the increased nutrients and elevated temperatures of the urbanized streams in Bellevue, as studied by the University of Washington as part of the U.S. EPA’s NURP project (EPA 1983). These conditions were probably responsible for accelerated growth of the fry which were observed to migrate to Puget Sound and the Pacific Ocean sooner than their counterparts in the control forested watershed that was also studied. However, the degradation of sediments, mainly the decreased particle sizes, adversely affected their spawning areas in streams that had become urbanized. Sovern and Washington (1997) reported that, in Western Washington, frequent high flow rates can be 10 to 100 times the predevelopment flows in urbanized areas, but that the low flows in the urban streams are commonly lower than the predevelopment low flows. They have concluded that the effects of urbanization on western Washington streams are dramatic, in most cases permanently changing the stream hydrologic balance by: increasing the annual water volume in the stream, increasing the volume and rate of storm flows, decreasing the low flows during dry periods, and increasing the sediment and pollutant discharges from the watershed. With urbanization, the streams increase in cross-sectional area to accommodate these increased flows and headwater downcutting occurs to decrease the channel gradient. The gradients of stable urban streams are often only about 1 to 2 percent, compared to 2 to 10 percent gradients in natural areas. These changes in width and the downcutting result in very different and changing stream conditions. The common pool/drop habitats are generally replaced by pool/riffle habitats, and the stream bed material is comprised of much finer material, for example. Along urban streams, fewer than 50 aquatic plant and animal species are usually found. They have concluded that once urbanization begins, the effects on stream shape are not completely reversible. Developing and maintaining quality aquatic life habitat, however, is possible under urban conditions, but it requires human intervention and it will not be the same as for forested watersheds.

Other Seattle area researchers have specifically examined the role that large woody debris (LWD) has in stabilizing the habitat in urban streams. Booth, et al. (1997) found that LWD performs key functions in undisturbed streams that drain lowland forested watersheds in western Washington. These important functions include: energy dissipation of the flow energy, channel bank and bed stabilization, sediment trapping, and pool formation. Urbanization typically results in the almost complete removal of this material. They point out that logs and other debris have long been removed from channels in urban areas for many reasons, especially because of their potential for blocking culverts or to form jams at bridges, they may increase bank scour, and many residents favor "neat" stream bank areas (a lack of woody debris in and near the water and even with mowed grass to the waters edge). Booth, et al. (1997) present and modify the stream classification system originally developed by Montgomery and Buffington (1993) that recognizes LWD as an important component of Pacific northwest streams that are being severely affected by urbanization. Figure 1 is a figure describing this general classification scheme.

Figure 1. Characteristics of channel types, after Montgomery and Buffington (1993).

The role of LWD varies in each channel type, and the effects of its removal also varies. The channel types are described as follows. The upper colluvial channels are wholly surrounded by colluvium (sediment transported by creep or landsliding, and not by stream transport) and generally lie at the top of the channel network. The cascade channels are the steepest of the alluvial channels and are characterized as having tumbling flows around individual boulders that dissipate most of the energy of the flowing water. Only very small pools are in cascade channels. The step-pool channels have accumulations of debris that form a series of steps that are one to four channel widths apart. The steps separate small pools that accumulate fine sediment. The fine sediment can be periodically flushed downstream during rare events. "Free" step-pool channels are characterized by steps that are made of alluvium that can be periodically transported downstream during high flows, while "forced" step-pool channels are characterized by steps that are made of immovable obstructions (large logs or bedrock). The removal of LWD from a forced step-pool stream in the Cascade Range could be naturally compensated by the common occurrence of large boulders that also form forced steps. However, in the lowlands near Puget Sound, the available sand and gravel stream deposits are too small to form stable steps, and the removal of LWD would have a much more severe effect on the channel stability. Plane-bed channels have long and channel-wide reaches of uniform riffles and do not have pronounced meanders and associated pools. Pool-riffle channels are the most common lowland stream channels in western Washington. These streams have pronounced meanders with pools at the outside of the bends and corresponding bars on the inside of the bends. Riffles form in the relatively straight stretch between the pools. There are also "free" and "forced" pool-riffle channels. Forced riffle-pool channels are typically formed with obstructions, such as LWD, and their removal would generally lead to a plane-bed channel characteristic. Forced riffle-pool channels form due to natural meanders and the inertial forces of the water. Dune-ripple channels have beds mostly made of sand where the character of the bed material changes in response to the flows.

The role of LWD is also highly dependent on the width of the stream. In narrow channels (high gradient colluvial and cascade channels), much of the LWD can be suspended above the flows, rarely being submerged and not available as a fish refuge, a sediment trap, or to dissipate the water’s energy. In wide channels (dune-ripple channels), the LWD may be significantly shorter than the channel width, with minimal stable opportunities to provide steps in the channel. Therefore, LWD plays a much more important role in channels having medium widths (lowland streams having plane-bed and pool-riffle channels), where the timber can become tightly lodged in the common flow channel. The removal of the LWD in these streams, especially in streams having few boulder steps, would have significant effects. Fish populations decline rapidly and precipitously following the removal of LWD in these critical streams (Booth, et al. 1997).

Horner, et al. (1997) described an extensive study in the Pacific Northwest where 31 stream reaches were examined since 1994 for a variety of in-stream and watershed characteristics. They felt that the most severe in-stream biological changes were most likely associated with changes in habitat, especially increased frequencies and magnitudes of high flows. These flow changes were therefore thought to most related to watershed factors affecting runoff, especially the amount of impervious areas in the watershed. Table XX4 shows the distribution of total impervious area (as a percentage of the watershed area) associated with each stream reach studied. They felt that the most rapid changes in ecological conditions were most likely to occur for urbanizing streams at relatively low levels of development, conditions representing most of the selected study sites. Table XX5 lists the variables examined during the stream reach studies.

Table XX4. Impervious Areas for Stream Reaches Examined by Horner, et al. (1997).

Table XX5. Variables Examined During Stream Reach Surveys (Horner, et al. 1997)

Figures 1 through 6 show resulting relationships between various indicators of stream biological conditions plotted against watershed total impervious area. Figure 1 shows a rapid decline in biological conditions as total imperviousness area increases to about 8% in the watershed. The rate of decline is less for higher levels of urbanization. Eight study areas (having B-IBI values of 25 to 31 and associated with impervious areas of 25 to 34%) indicate better biological conditions than expected. These areas were associated with higher amounts of intact wetlands along the riparian corridors than other sites, indicating a possible significant moderating effect associated with preserving stream corridors in their natural condition. Figure 2 illustrates that the less tolerant coho salmon is much more abundant than the more tolerant cutthroat trout only for very low levels of urbanization. Stormwater concentrations of zinc are seen to increase steadily with increasing impervious areas in Figure 3. However, the concentrations are well below the critical water quality criteria until the impervious cover reaches about 40%, a level much greater than when significant biological effects are noted. Similar conclusions were made with other water pollutant metal concentrations and pollutant concentrations in the sediment. Horner, et al. (1997) interpreted these findings to imply that pollutant conditions were much less important than habitat destruction when affecting in-stream biological conditions. Figure 5 shows distinct relations between increased hydrologic responses and urbanization. Figure 6 plots the amount of large woody debris with urbanization, also showing significant decreases in this important habitat component with increasing development.

They concluded that the preponderance of physical and biological evidence indicated rapid in-stream biological conditions at early stages of urbanization. However, chemical pollutants did not appear to significantly affect biological conditions in the early stages of urbanization, but may have at very high levels of urbanization. Based on their results, they developed a preliminary summary of the conditions that would allow high levels of biological functions in the Puget Sound area:

Figure 1. Benthic Index of Biotic Integrity over a gradient of total impervious area (Horner, et al. 1997).

Figure 2. Juvenile coho salmon/cutthroat trout over a gradient of total impervious area (Horner, et al. 1997).

Figure 3. Zinc concentrations in storm flows over a gradient of total impervious area (Horner, et al. 1997).

Figure 5. Relative 2-year frequency flood rise over a gradient of total impervious area (Horner, et al. 1997).

Figure 6. Large woody debris quantity over a gradient of total impervious area (Horner, et al. 1997).

Habitat evaluations are commonly and justifiably recognized as critical components of stream and watershed studies. However, Poole, et al. (1997) warn against their use to quantify aquatic habitat or channel morphology in an attempt to measure the response of individual streams to human activities. Their concern is the subjectivity of habitat surveys and the lack of repeatability, precision, and transferability of the measurement techniques. The measurement parameters are also assigned relatively arbitrary nominal values that are not easily statistically evaluated. They feel that the typical use of habitat unit classifications encourages the focus on direct manipulation or replacement of habitat structures (such as in stream "restoration" activities, described in Chapter 13) while neglecting the long-term maintenance of habitat-forming biophysical processes (such as controlling the energy distribution of stream discharges and the discharges of sediment into the streams).

Figure 2 is an example of the lack of repeatability of habitat surveys from two sets of observations conducted within a day by two trained teams. This most basic classification effort shows major inconsistencies, implying a shift from pools and cascades to riffles. They found that different trained observers were often unable to consistently distinguish between unit types of habitat at levels significantly better than random classifications. Poole, et al. (1997) calculated the 95% confidence intervals for the habitat unit classifications from prior studies on a number of different streams. They found that the confidence intervals ranged from ± 5.6% to ± 20.7% of the mean for pool areas and ± 5.5 to ± 20.4% of the mean for riffle areas. It was also found that a more highly trained and experienced group of observers did not show a consistent improvement over a less experienced group. They concluded that reported apparent changes in habitat unit types may be relatively insensitive to actual conditions.

Figure 2. Maps of the same stream each based on two surveys conducted within a 24-hr period (Poole, et al. 1997).

Therefore, the use of habitat unit classifications as an indicator of watershed health may be most appropriately used for only very large differences or changes, when conducted over a large portion of a watershed being studied, and only if a sufficiently large number of observations and replicates are made to compensate for the high inherent measurement variations. Many current habitat surveys are being conducted on small scales within a short period of time and with few observations, and without adequate statistical evaluations of the data. The results of these surveys are therefore of questionable value. As for all indicators, it is important that methods be developed and tested to improve the accuracy of the tool, and that additional supplemental measurement methods also be used to confirm observations and conclusions, especially when evaluating cause and effect relationships in watersheds.

Many of the observed biological effects associated with urban runoff may be caused by polluted sediments and associated benthic organism impacts. The EPA (1998) prepared a four volume report to Congress on the incidence and severity of sediment contamination in the surface waters of the U.S. This report was required by the Water Resources Development Act of 1992. This Act defines contaminated sediment as "sediment containing chemical substances in excess of appropriate geochemical, toxicological or sediment quality criteria or measures; or otherwise considered to pose a threat to human health or the environment." Volume 1 of this report is the National Sediment Quality Survey and is a screening analysis to qualitatively assess the probability of associated adverse human or ecological effects at sampling stations based on a weight-of-evidence evaluation. Volume 2 is the Data Summary for Areas of Probable Concern (APC) and includes sampling station location maps, and chemical and biological summary data for the APC watersheds. Volume 3 is the National Sediment Contaminant Point Source Inventory and is a screening analysis to identify probable point source contributions of sediment pollutants. Volume 4 (under development) is the National Sediment Contaminant Nonpoint Source Inventory and is a screening analysis to identify probable nonpoint sources of contaminated sediment. In the national quality survey, the EPA examined data from 65% of the 2,111 watersheds in the U.S. and identified 96 watersheds that contain areas of probable concern. In portions of these waters, benthic organisms and fish may contain chemicals at levels unsafe for regular consumption. Areas of probable concern are located in regions affected by urban and agricultural runoff, municipal and industrial waste discharges, and other pollutant sources. When the fourth volume is completed, much more detailed information will become available concerning the relative role that urban stormwater contributes to national contaminated sediment problems.

Examples of heavy metal and nutrient accumulations in urban sediments are currently numerous. DePinto, et al. (1980) found that the cadmium content of river sediments can be more than 1,000 times greater than the overlying water concentrations and the accumulation factors in sediments are closely correlated with sediment organic content. They reported that sediments were also able to adsorb phosphorus in proportion to the phosphorus concentrations in the overlaying waters during aerobic periods, but that the sediments released phosphorus during anaerobic periods. Heaney (1978) found that long-term impacts of urban runoff related to the resuspension of previously deposited polluted benthos material may be more important than short-term discharges of pollutants from potential "first-flushes."

Another comprehensive study on polluted sediment was conducted by Wilber and Hunter (1980) along the Saddle River in New Jersey where they found significant increases in sediment contamination with increasing urbanization. They found large variations in metal concentrations for different sediment particle sizes in the urban river. The sediment particle size distribution was the predominant influencing factor for total metal concentrations in the sediments. Areas having fine sediments had a substantially greater concentration of heavy metals than those areas having coarse sediments.

Table XXXD Wilbur and Hunter data

In another study, Pitt and Bozeman (1982) observed concentrations for many contaminants in the urban area sediments of Coyote Creek (San Jose, California) that were much greater than those from the nonurban area. This data is summarized in the case study presented at the end of this chapter. Orthophosphates, TOC, BOD₅, sulfates, sulfur, and lead were all found in higher concentrations in the sediments from the urban area stations, as compared with those from the upstream, non-urban area stations. The median sediment particle sizes were also found to be significantly smaller at the urban area stations, reflecting a higher silt content.

Several of the University of Washington projects and the Seattle METRO project investigated physical and chemical characteristics of the Kelsey and Bear Creeks sediments as part of the Bellevue, WA, NURP projects (Pitt and Bissonnette 1984). Perkins (1982) found that the size and composition of the sediments near the water interface tended to be more variable and of a larger median size in Kelsey Creek than in Bear Creek. These particle sizes varied in both streams on an annual cycle in response to runoff events. Larger particle sizes were more common during the winter months when the larger flows were probably more efficient in flushing through the finer materials. Pedersen (1981) also states that Kelsey Creek demonstrated a much greater accumulation of sandy sediments in the early spring. This decreases the suitability of the stream substrates for benthic colonization. Scott, et al. (1982) state that the level of fines in the sediment samples appears to be a more sensitive measure of substrate quality than the geometric mean of the particle size distribution. Fines were defined as all material less than about 840 microns in diameter. METRO (Galvin and Moore 1982) also analyzed organic priority pollutants in 17 creek sediments including several in Kelsey and Bear Creeks. Very few organic compounds were detected in either stream with the most notable trend being the much more common occurrence of various PAHs in Kelsey Creek while none were detected in Bear Creek.

Scott, et al. (1982) state that streambed substrate quality can be an important factor in the survival of salmonid embryos. Richey (1982) describes sediment bioassay tests which were performed using Kelsey and Bear Creeks sediments. She found that during the four-day bioassay experiment, no mortalities or loss of activities were observed in any of the tests. She concluded that the chemical constituents in the sediment were not acutely toxic to the test organism. However, the chronic and/or low level toxicities of these materials was not tested.

The University of Washington project and the Seattle METRO project analyzed interstitial water for various constituents. These samples were obtained by inserting perforated aluminum stand pipes into the creek sediment. This water is most affected by the sediment quality and affects in turn the benthic organisms much more than the creek water column. Scott, et al. (1982) found that the interstitial water pH ranged from 6.5 to 7.6 and did not significantly differ between the two streams but did tend to decrease during the spring months. The lower fall temperatures and pH levels contributed to reductions in ammonium concentrations. The total ammonia and ammonium concentrations were significantly greater in Kelsey Creek than in Bear Creek. They also found that the interstitial dissolved oxygen concentrations in Kelsey Creek were much below those concentrations considered normal for undisturbed watersheds. These decreased interstitial oxygen concentrations were much less than the water column concentrations and indicated the possible impact of urban development. The dissolved oxygen concentrations in the interstitial waters and Bear Creek were also lower than expected potentially suggesting deteriorating fish spawning conditions. During the winter and spring months, the interstitial oxygen concentrations appeared to be intermediate between those characteristic of disturbed and undisturbed watersheds.

The University of Washington (Richey 1982) also analyzed heavy metals in the interstitial waters. They found that copper and chromium concentrations were very low or undetectable, while lead and zinc were higher. Kelsey Creek interstitial water also had concentrations approximately twice those found in the Bear Creek interstitial water. They expect that most of the metals were loosely bound to fine sediment particles. Most of the lead found was associated with the particulates and very little soluble lead was found in the interstitial waters. The interstitial samples taken from the stand pipes were full of sediment particles which could be expected to release lead into solution following the mild acid digestion for exchangeable lead analyses. They also found that the metal concentrations in Kelsey Creek interstitial water decreased in a downstream direction. They felt that this might be caused by stream scouring of the benthic material in that part of the creek. The downstream Kelsey Creek sites were more prone to erosion and channel scouring while the most upstream station was relatively stable.

Seattle METRO (Galvin and Moore 1982) also monitored heavy metals in the interstitial waters in Kelsey and Bear Creeks. They found large variations in heavy metal concentrations depending upon whether the sample was obtained during the wet or the dry season. During storm periods, the interstitial water and creek water heavy metal concentrations approached the stormwater values (200 m g/L for lead). During non-storm periods, the interstitial lead concentrations were typically only about 1 m g/L. They also analyzed priority pollutant organics in interstitial waters. Only benzene was found and only in the urban stream. The observed benzene concentrations in two Kelsey Creek samples were 22 and 24 m g/L, while the reported concentrations were less than 1 m g/L in all other interstitial water samples analyzed for benzene.

The effects of erosion and sediment deposition in urbanized Kelsey Creek were more severe than found in non-urbanized Bear Creek. Kelsey Creek’s channel was scoured to a deeper depth, there was much more channel instability in Kelsey Creek, and the numbers of erosion cuts and deposition bars were much more frequent in Kelsey Creek. Richey (1982) reported that the sediment transport in Bear Creek during December of 1979 was 27 kg/ha while 98 kg/ha left Kelsey Creek. The suspended solids transport was almost exponentially related to discharge. On an annual basis, Kelsey Creek discharged almost four times as much suspended solids as did Bear Creek, but most of this material passed through the stream in a few hours or days. Richey (1982) found that much of the solids transport in Kelsey Creek occurred during the rapid rise of the hydrograph when the energy to move sediment material was increasing. The silts and associated pollutants were rapidly transported through the system during these periods. The scouring of the channel appeared to remove temporarily stored silts and the associated pollutants. The higher levels of particulate transport in Kelsey Creek are probably because of increased stream power rather than because of increased sources of sediment material in the watershed. However, there were substantial amounts of in-stream sources of sediment material in Kelsey Creek to augment the runoff discharged sediment. Because of the lack of debris dams in the downstream sections of Kelsey Creek, the transported materials are carried significant distances before deposition. The high stream power available to transport the materials and the erodable nature of the stream banks in the watershed areas along with the lack of storage sites along the stream all contributed to high particulate yields from Kelsey Creek. Because much of the suspended particulate material in Kelsey Creek was from the relatively unpolluted bank materials, the sediments and suspended loads in Kelsey Creek had much lower concentrations of many of the typical urban pollutants compared to the urban runoff that was discharged to the creek.

A number of recent investigations have examined sediment quality in conjunction with biological conditions in urban receiving waters in attempts to identify causative agents affecting the biological community. Arhelger, et al. (1996) examined conditions in the upper Houston Ship Channel that receives drainage from the metropolitan Houston area. The channel has been dredged to allow large vessels access to the upper reaches of what used to be a relatively small channel. The dredging has increased the cross-sectional area by about 20 times, with attendant significant decreases in flushing flows. This has allowed efficient sedimentation of suspended material discharged from the 500 mi² urban watershed. The sediments have undergone extensive chemical, physical, and toxicity testing, with frequent indications of toxicity. The tests have indicated that the toxicity is most likely caused by the high sediment oxygen demand and associated low dissolved oxygen conditions. Toxicity testing of Ampelisca under varied DO conditions showed significant decreases in survival when the bottom DO is less than 3 mg/L, for example. Even though the point source BOD loads have been reduced by more than 90% since the 1970s, receiving water and sediment oxygen levels are very low, presumably caused by uncontrolled stormwater sources.

Previous studies near Auckland, New Zealand have shown that sediment concentrations of many constituents near stormwater outfalls, especially in industrial areas, often exceed guidelines intended to protect bottom-dwelling animals. Guidelines used were as presented by Long, et al. (1996) and were as follows (along with sediment concentrations from two locations near Auckland):

Lead, zinc, and organochlorides were the most widespread potential problems. Field surveys and laboratory toxicity tests had shown circumstantial evidence of chronic toxicity associated with stormwater. Detailed field surveys, by Morrisey, et al. (1997), were therefore conducted to better understand actual toxicity problems in the local marine estuaries that are influenced by complex natural factors. These complicating factors include strong gradients in salinity, sediment texture, currents, and wave action, all radically affecting the natural distribution of benthic fauna. In slowly growing areas or in relatively low density urban areas, the relatively small rate of accumulation of contaminated sediments from nonpoint sources may take many years to accumulate to levels that may produce detectable impacts in the receiving waters. In addition, changing urban conditions and changing weather from year to year make the rate of accumulation highly variable. These factors all make it difficult to conduct many types of field experiments that rely on before and after observations, or other short-term observations that assume steady conditions. They therefore relied on a "weight-of-evidence" approach considering many different and reinforcing/confirming procedures (such as the sediment quality triad and the effects range tests, both of which rely on distribution of contaminants and organisms in the field and from laboratory toxicity tests). They also applied their results to the Abundance Biomass Comparison index proposed by Warwick (1986). This index is a relative measure of biomass vs. abundance and has been shown to work well for individual sites where control sites are difficult to identify and study, especially if available "control" sites already impacted. Pore water chemistry, sediment quality, and benthic community composition were included in the field analyses. Statistical analyses identified the strongest correlations between pH and iron content of the pore water and the sediment texture, with benthic composition. The pH and iron pore water conditions may affect the bioavailability of the sediment heavy metals. Current and future work includes similar studies in non-urbanized estuaries, the development of chronic toxicity tests using local indigenous organisms, and studies of recolonization of heavily impacted sites. .

Watzin, et al. (1997) examined sediment contamination in Lake Champlain near Burlington, VT, to compare several toxicity endpoints with sediment characteristics. They measured sediment pore water toxicity using Ceriodaphania dubia, Chironomus tentans, and Pimephales promelas, benthic community composition, and many physical and chemical characteristics at 19 locations. Four major storm drains and the secondary sewage treatment plant all discharged to the harbor. Boat traffic and historical petroleum handling facilities also affected some of the sampling locations. They found variable levels of toxicity at the different sites, but effects of acid-volatile sulfides on heavy metal toxicity was not demonstrated. However, they did find strong associations between metal and organic carbon levels and toxicity, indicating possible metal-organic matter complexation reducing metal availability. The sediment toxicity tests did indicate a moderate level of concern, but the macroinvertebrate community was apparently not significantly affected during these tests. They propose the use of a weight-of-evidence approach that uses multiple indicators of problems and possible sources of the problems, plus repeated observations over seasonal cycles, before management recommendations are developed.

Equilibrium partitioning of sediment-based fluoranthene and critical bioaccumulation levels was used to predict toxicity to amphipods by Driscoll and Landrum (1997). The equilibrium partitioning theorey (EqP) has been used to predict effects of organic toxicants found in sediments, using an organic carbon-normalized sediment concentration of the hydrophobic organic compound (used for PAHs and pesticides) and resulting estimated pore-water concentrations. They report that toxicity bioassays with benthic invertebrates have, in general, confirmed this approach. However, certain test organisms and sediments have not been well predicted using this approach. Driscoll and Landrum tested a complementary method: the critical body residue (CBR) approach. This method measures the actual body burdens of a compound in relation to toxic effects. They found that the CBR approach is a useful complement to the EqP approach for the prediction and assessment of toxicity associated with contaminated sediments.

The effects of large discharges of relatively uncontaminated sediment on the receiving water aquatic environment were summarized by Schueler (1997a). These large discharges are mostly associated with poorly controlled construction sites, where 30 to 300 tons of sediment per acre per year of exposure may be lost. These high rates can be 20 to 2,000 times the unit area rates associated with other land uses. Unfortunately, much of this sediment reaches urban receiving waters, where massive impacts on the aquatic environment can result. Unfortunately, high rates of sediment loss can also be associated with later phases of urbanization, where receiving water channel banks widen to accommodate the increased runoff volume and frequency of high erosive flow rates. Sediment is typically listed as one of the most important pollutants causing receiving water problems in the nations waters. Schueler (1997a) listed the impacts that can be associated with suspended sediment:

"· abrades and damages fish gills, increasing risk of infection and disease

· scouring of periphyton from streams (plants attached to rocks)

· loss of sensitive or threatened fish species when turbidity exceeds 25 NTU

· shifts in fish communities toward more sediment tolerant species

· decline in sunfish, bass, chub, and catfish when monthly turbidity exceed 100 NTU

· reduces sight distance for trout, with reduction in feeding efficiency

· reduces light penetration that causes reduction in plankton and aquatic plant growth

· reduces filtration efficiency of zooplankton in lakes and estuaries

· adversely impacts aquatic insects which are the base of the food chain

· slightly increases stream temperature in summer

· suspended sediments are a major carrier of nutrients and metals

· turbidity increases probability of boating, swimming, and diving accidents

· increased water treatment to meet drinking water standards

· increased wear and tear on hydroelectric and water intake equipment

· reduces anglers chances of catching fish

· diminishes direct and indirect recreational experience of receiving waters"

He also listed the impacts that can be associated with deposited sediment:

"· physical smothering of benthic aquatic insect community

· reduced survival rates for fish eggs

· destruction of fish spawning areas and redds

· ‘imbedding’ of stream bottom reduces fish and macroinvertebrate habitat value

· loss of trout habitat when fine sediments are deposited in spawning or riffle-runs

· sensitive or threatened darters and dace may be eliminated from fish community

· increase in sediment oxygen demand can deplete DO in lakes or streams

· significant contributing factor in the alarming decline of freshwater mussels

· reduced channel capacity, exacerbating downstream bank erosion and flooding

· reduced flood transport capacity under bridges and through culverts

· loss of storage and lower design life for reservoirs, impoundments, and ponds

· dredging costs to maintain navigable channels and reservoir capacity

· spoiling of sand beaches

· deposits diminish the scenic and recreational value of waterways"

Sediment Contamination Effects and Criteria. There is much concern and discussion about contaminated sediments in urban receiving waters. Many historical discussions downplayed the significance of contaminated sediments, based on their assumed "low-availability" to aquatic organisms. However, many of the previously described receiving water studies found greatly disturbed benthic organism populations at sites with contaminated urban sediments, compared to uncontaminated control sites. More specifically, in-situ sediment toxicity tests in urban receiving waters (such as those conducted by Burton and Stemmer 1988; Burton 1989, 1991, and 1992; Burton, et al. 1989, Burton and Scott 1992; and Crunkilton, et al. 1997) have illustrated the direct toxic effects associated with exposure to contaminated urban sediments, to problems associated with their scour, and to decreases in toxicity associated with their removal from stormwater.

The fate of contaminated sediments, especially mechanisms that expose contaminants to sensitive organisms, can determine the overall and varied effects that the sediments may have. Scour of fine-grained sediments during periods of high flows in streams and rivers, or due to turbulence from watercraft in shallow waterbodies, has frequently been encountered. In addition, pollutant remobilization may also occur through bioturbation from sediment-dwelling microorganisms, or from nest-building fish. These mechanisms may resuspend contaminants, making them more available to organisms. Burrowing organisms can also transport deeply buried contaminants to surface layers, thereby increasing surface contamination levels, while the surface scouring mechanisms would tend to decrease the concentrations in the surface sediment. Bioturbation has been reported to strongly influence the fate of contaminants and that sediment-bound contaminants can be remobilized by biological activity (ES&T 1997).

Lee and Jones-Lee (1996) reviewed the significance of chemically contaminated sediments and associated impacts. They are especially concerned about the development of sediment contamination criteria based on simple chemical tests. They feel that is has been well demonstrated that the toxic-available form of chemical constituents present in the sediment is the dissolved form present in the interstitial waters. Historically, the EPA assumed that the dissolved form of certain organic toxicants could be estimated based on an equilibrium partitioning model based on the particulate organic carbon present. Likewise, the dissolved forms of heavy metals were assumed to be controlled by metal sulfide precipitates. Lee and Jones-Lee feel that the EPA’s overly simplistic two component box model used to predict dissolved forms of toxicants should never be used alone without concurrent well-established toxicity measurements. They are also concerned about the use of toxicity co-occurrence data bases used to relate measured sediment chemical conditions with observed biological conditions that are also sometimes used to establish sediment criteria. These data bases have not considered some of the most important possible causes of toxicity at the test sites, namely low dissolved oxygen, and high ammonia and hydrogen sulfide concentrations. They outlined the components of sediment toxicity tests that they feel are necessary:

They feel that the best approach in developing sediment quality evaluations should use a best professional judgement (BPJ), weight-of-evidence approach. This approach involves an integrated assessment of the aquatic life toxicity test results, assessment of the bioaccumulations of hazardous chemicals in edible portions of aquatic life, knowledge of chemical characteristics of the sediments and associated waters, and investigations of the aquatic life assemblages in the sediments of concern compared to appropriate reference sites.

Kuehne (1975) studied the usefulness of using various aquatic organisms during stream taxonomic surveys as indicators of pollution. He found that invertebrates can reveal pollution for some time after a water pollution event, but they cannot give accurate indications of the nature of the pollutants. He stated that in-stream fish studies had not been employed as biological indicators much before 1975, but that they are comparable in many ways to invertebrates as quality indicators and can be more easily identified. However, because of better information pertaining to invertebrates and due to their limited mobility, certain species may be useful as sensitive indicators of minor changes in water quality. Fish can be highly mobile and cover large sections of a stream, as long as their passage is not totally blocked by adverse conditions. Fish disease surveys were also used during the Bellevue, Washington, urban runoff studies as an indicator of water quality problems (Scott, et al. 1982; Pitt and Bissonnette 1984). McHardy, et al. (1985) also examined heavy metal uptake in green algae (Cladophora glomerata) from urban runoff for use as a biological monitor of specific metals.

Burton and Stemmer (1988), during tests conducted at polluted stream and landfill sites, found that a battery of laboratory and in-situ bioassay tests were most useful when determining aquatic biota problems. The test series included microbial activity tests, along with exposures of microfaunal organisms, zooplankton, amphipods, and fathead minnows to the test water. The newly developed microbial tests correlated well with in-situ biological test results. Bascombe, et al. (1990) also reported on the use of in-situ biological tests, using an amphipod exposed for five to six weeks in urban streams, to examine urban runoff receiving water effects. Ellis, et al. (1991) examined bioassay procedures for evaluating urban runoff effects on receiving water biota. They concluded that an acceptable criteria for protecting receiving water organisms should not only provide information on concentration and exposure relationships for in-situ bioassays, but also consider body burdens, recovery rates, and sediment related effects.

A number of stormwater researchers have recently presented bioassessment and other "watershed indicators" that they have found as useful tools to quantify local receiving water problems. Many of these schemes were presented at the Assessing the Cumulative Impacts of Watershed Development in Aquatic Ecosystems and Water Quality conference held in Chicago in March of 1996, sponsored by the Northeastern Illinois Planning Commission, and at the Effects of Watershed Development and Management on Aquatic Ecosystems conference held in Snowbird, UT, in August of 1996, sponsored by the Engineering Foundation and the ASCE. Several papers from those conferences are summarized below, by location.

Barbour (1997) reviewed many of the state programs throughout the U.S. that are using biological assessments as part of their water resources programs. Most of the active state bioassessment programs started since 1990, after the publication of the EPA’s Rapid Bioassessment Protocols (Plafkin, et al. 1989) and the Program Guidance for Biocriteria (EPA 1990) manuals. By 1996, numeric biocriteria were in place in Ohio and Florida (and promulgated in Maine) and under development in 13 other states. Although the majority of the states had not used biocriteria, nearly ¾ had used bioassessment data to measure the attainment of their aquatic uses. Almost all states were using benthic macroinvertebrates (all but 3 states) and fish (all but 14 states). Seven states were also using algae in their bioassessment programs.

An important aspect of the biocriteria approach is that local and regional expectations be considered in setting specific objectives. In addition, local reference sites representing specific ecoregions are also used to calibrate observations. The basic components of a bioassessment include:

The following is a summary of some of the successful state programs that are using bioassessments as part of their local programs. As shown, elements of the specific programs and the uses of the data vary.

Maxted (1997) examined the effects of urbanization on small (first and second order) perennial streams in Delaware using macroinvertebrate and habitat metrics. These small streams make up about 85% of the nontidal streams in the state, but have received little attention because of their small size and lack of major point source discharges. It was felt that aquatic organism (especially macroinvertebrates) and habitat assessments would be much more cost effective than conventional water quality and flow monitoring for measuring the relative effects of stormwater on the different streams. Table 1 shows the biological and habitat metrics used at 38 sites during the fall of 1993.

Macroinvertebrate samples were taken with a 1m² kick net at each site in riffle habitats and 100 organisms were identified to the genus level. This information was used to calculate each of six biological metrics. The habitat scores were determined by comparisons with reference sites. Land use information and imperviousness data were then compared to these metrics, as shown in Figures XXX1 through 3. These plots indicate that habitat and the macroinvertebrate indices were related: it was very unlikely that good quality macroinvertebrate communities could exist in areas of poor habitat. In addition, the streams in the urban areas were observed to have specific characteristics: eroded stream banks along both bends and runs, uniform and shallower depth, wider channels, and newly deposited sediment in the channels, all leading to degraded habitat conditions. Biological quality was found to decrease by about 50% compared to the reference sites, after the percent imperviousness (based on NRCS estimates for each land use) reached about 10 to 15%. About 90% of the sensitive macroinvertebrates (mayflies stoneflies, and caddisflies) were lost from the urban streams at this level of development, being replaced by macroinvertebrates from the more tolerant chironomidea family. Statistical analyses showed that the more urbanized watersheds (from 15 to 50% imperviousness) did not vary greatly in their biological quality, indicating that a threshold condition may have been reached.

Figures XXX1 through 3.

McCarron, et al. (1997) and Livingston, et al. (1997) described the comprehensive bioassessment program developed and used in Florida since 1990. They have concluded that biological community monitoring is an important component of this assessment program because of its ability to detect both cumulative and episodic pollutant discharges and the effects of habitat changes. Florida’s program includes many of the concepts of the EPA’s Rapid Bioassessment Protocols for use in Steams and Rivers (Plafkin, et al. 1989). These protocols stress the need for a broad set of assessment tools, including habitat effects associated with stream flow and physical structure and biological measurements, using local reference cases for comparison. Florida’s program relies greatly on macroinvertebrate evaluations for numerous reasons (including their limited migration, their sensitivity to short-term problems, their relative ease and low cost to sample, and their abundance in most streams).

The first step for Florida was to identify and investigate relatively undisturbed reference stream locations. From 6 to 13 reference sites were finally examined in each of 9 subecoregions, totaling 83 reference locations throughout the state. They found that it is important to have a relatively large number of reference sites because of long-term watershed changes that were likely to affect each site, and because of the variable existing level of development throughout the state. The reference sites are not expected to be completely natural, as that is impossible in a rapidly changing state such as Florida. However, they should represent the least disturbed conditions available in the subecoregion. The characteristics of the reference locations were used to calibrate the biological monitoring conducted at disturbed locations in each subecoregion.

Florida’s Stream Condition Index (SCI) is an aggregate index, calculated using different biological metrics, all based on stream macroinvertebrate community evaluations. The SCI is used to indicate the general biological health of a stream location, compared to the conditions at an appropriate reference location in the same subecoregion. Florida evaluated 47 biological metrics that were thought to be relevant to local conditions. The metrics represented "richness" (relative diversity of aquatic organisms), "composition" (make-up and relative abundance of taxa), "tolerance" (sensitivity of components of the biological community), and "trophic conditions" (surrogates for complexity of food structure). Florida selected a subset of the 47 metrics evaluated that were independent measures of the impacts, as shown in Table 1.

A less rigorous stream assessment than the SCI is the bioreconnaissance (BioRecon) which can be used as a screening tool before a SCI is determined, or to provide a rapid assessment of conditions at the least cost. The biological survey that is part of the BioRecon requires substantially less time than a full SCI determination. McCarron, et al. (1997) estimated that about 5 hours was needed to conduct the field collection and lab sorting for the BioRecon, while more than 30 hours was required to determine a SCI score for a site.

A habitat evaluation is an important component for either an BioRecon or a SCI-based survey. The habitat evaluation includes both physical/chemical characterizations and a habitat assessment. The physical/chemical characterization identifies predominant land use, local erosion, pollution sources, stream depth and width, high water mark, temperature, and water velocity. Selected water quality parameters are also measured (pH, DO, conductivity, transparency, water clarity, color, odors and the presence of surface oils, etc.). The habitat assessment evaluated the water velocity, substrate and cover, channel conditions, bank stability, and riparian vegetation. Figure 1 shows the habitat assessment field data sheet used in Florida.

Figure 1. Habitat Assessment Field Data Sheet (McCarron, et al. 1997).

Dreher (1997) described the correlation of the Index of Biotic Integrity (IBI) in Illinois to different levels of urbanization. Over the past 20 years, point source discharges have dramatically been reduced in the urbanized areas of Illinois, with significant reductions in pollutants associated with municipal point sources and CSOs. However, Illinois’ most recent statewide water quality assessment shows that many of the regions waterways are still seriously impaired. The current impairments were most likely associated with stormwater discharges, although some point source and CSO discharges are also problematic. It was therefore desired to see if there was a relationship between the extent of urbanization and stream degradation. They selected measurements of fish diversity and abundance as their indicator of overall stream quality.

A cause-and-effect relationship between urbanization and stream use was strongly implied. Existing questions posed by Illinois included: could extensive stormwater controls in newly developing areas affect this relationship, and can retro-fitted controls allow the recovery of some of the streams’ uses? Physical habitat and the presence of a re-colonizing fish population were assumed to be critical. This assessment procedure relied solely on fish information, as it was assumed that "fish reflect the full range of environmental influences, including water quality, sediment quality, and physical habitat – and they have been identified as useful, or even ‘ideal’ environmental indicators." Fish were also selected because they can be quantitatively assessed.

Galli (1997) described the Rapid Stream Assessment Technique (RSAT) developed and used in Maryland. The RSAT was developed by the Metropolitan Washington Council of Governments to identify channel erosion problem areas and to systematically characterize stream quality conditions. RSAT is based on both the EPA’s Rapid Bioassessment Protocols (Plafkin, et al. 1989), plus local previous stream survey work. It includes the following six components:

· channel stability,

· channel scouring/sediment deposition,

· physical aquatic habitat,

· water quality,

· riparian habitat conditions, and

· biological indicators.

RSAT also uses a reference stream approach for local calibration. RSAT uses over 30 physical, chemical and biological parameters at each section, all stream riffle stations, usually about 400 ft apart. The stream in its entirety is surveyed using RSAT, from upstream to downstream locations. An experienced 2-person team can survey about 1 to 1-¼ stream miles a day (about 12 to 15 stream transects). An example RSAT stream survey form is presented in Figures 1 and 2, showing the physical conditions of the stream, along with the water quality and biological conditions.

Figure 1. Example RSAT stream survey form (Galli 1997).

Stream channel stability and cross-sectional characterization is a critical component of RSAT. Either a 50 or 100-ft long channel reach is surveyed at each transect. Signs of instability, (such as bank sloughing, recently exposed non-woody tree roots, general absence of vegetation within bottom 1/3 of the bank, recent tree falls, etc.), and channel degradation or downcutting (such as high bank heights in small headwater streams and erosion around man-made structures) are noted, and cross-section measurements are made. Figure 2 is an example of a stream cross section in a heavily urbanized watershed, compared to a comparable stream in a less developed watershed. The less developed stream’s cross-sectional area is only about 20% of the cross-sectional area of the more urbanized stream reach, indicating downcutting and widening in response to increased stream flows.

Figure 2. Comparison of stream cross-sections of a heavily urbanized stream with a less urbanized stream (Galli 1997).

Soil sampling along the stream banks is also conducted to determine soil texture and potential erodibility of the stream bank. Limited water quality measurements are also made (temperature, DO, pH, conductivity, turbidity, TDS, color and odor), along with an indication of substrate fouling. The majority of the stream work involves macroinvertebrate sampling and evaluations. Macroinvertebrates associated with 10 cobble-sized stones, or larger, are examined, along with at least three 1 ft² (30 s) kick net samples at each transect. A macroinvertebrate community score is calculated based on the observations. Figure 3 shows an example plot of the macroinvertebrate community score along a stream, as it enters more urbanized areas of the watershed. The upper reaches of the stream are in a semi-rural area and it then flows into an older residential area.

Figure 3. Macroinvertebrate community score along a stream ((Galli 1997).

The RSAT has also been used to evaluate conditions above and below areas where stream improvement efforts and other stormwater controls have been used.

The Ohio EPA has often been recognized for having one of the more advanced biological assessments in place, especially in their efforts in incorporating biological criteria as part of their regulatory program. Yoder and Rankin (1997) reported that biological monitoring of small streams in Ohio has indicated a general lowering of biological index scores with increasing urbanization, especially in areas having CSOs and industrial discharges. Of 110 sampling sites investigated, only 23% had good to exceptional biological resources. Poor or very poor scores were evident in 85% of the urbanized areas. They also found that more than 40% of the suburban, urbanizing, sites were impaired, due to increasing new residential and commercial developments. An earlier Ohio study found that biological impairments were evident in about half of the locations where no impairments were indicated, based on chemical ambient monitoring data alone. They have therefore come to rely on biological monitoring, such as expressed in the Index of Biotic Integrity (IBI) and the Invertebrate Community Index (ICI), as a less expensive and as a more accurate overall indication of receiving water problems than conventional chemical water pollutant monitoring.

Figure 5 is a plot showing the attainability of receiving water uses in Ohio, based on the watershed development characteristics, as measured using biological monitoring (IBI scores). Yoder and Rankin (1997) show that the lower IBI scores and a subsequent loss of biological integrity occurs with an increasing degree of urbanization, especially in the presence of other associated urban impacts (such as CSOs, and industrial and commercial development). They also found that the severity of biological impairments in urban areas was influenced by steam and river size (indicated by watershed area), with the most severe effects in small headwater streams (watersheds of less than 20 mi² in area).

Figure 5. Frequency of biological monitoring locations in Ohio that do not attain receiving water uses (Yoder and Rankin 1997).

Yoder and Rankin (1997) found that urban stream sediment concentrations of heavy metals are commonly elevated compared to reference sites. Figure 6 shows relations between IBI and ICI scores and concentrations of heavy metals (arsenic, cadmium, chromium, copper, lead, nickel, and zinc). As the concentrations increase, the biological integrity decreases, with very few sites meeting the biological criteria at high sediment concentration sites. They concluded that sediment heavy metal levels reflect the history of heavy metal loadings from all sources. This history cannot be determined through typical spotty water quality monitoring. However, nontoxic effects of urbanization are also significant at many locations, especially siltation of stream bottoms associated with watershed erosion, identified as the second leading cause of impairment in Ohio streams in the 1994 Ohio Water Resource inventory. Sedimentation surpassed ammonia and heavy metals on the list of significant detrimental sources between the 1988 and 1994 surveys, and remains behind the number one impairment source, organic enrichment/DO problems (associated with both agricultural and urban uses).

Figure 6. Correlation between sediment heavy metal concentrations and biocriteria attainment (Yoder and Rankin 1997).

The EPA (1996) published a list of 18 indicators to track the health of the nation’s aquatic ecosystems. These indicators are intended to supplement conventional water quality analyses in compliance monitoring activities. The use of broader indicators of environmental health is increasing. As an example, 12 states are currently using biological indicators, and 27 states are developing local biological indicators, according to Pelley (1996). Because of the broad nature of the nation’s potential receiving water problems, this list is more general than typically used for specific stormwater issues. These 18 indicators are (EPA 1996):

1) population served by drinking water systems violating health-based requirements.

2) population served by unfiltered surface water systems at risk from microbiological contamination.

3) population served by communities by community drinking water systems exceeding lead action levels.

4) drinking water systems with source water protection programs.

5) fish consumption advisories.

6) shellfish-growing waters approved for harvest for human consumption.

7) biological integrity of rivers and estuaries.

8) species at risk of extinction.

9) rate of wetland acreage loss.

10) designated uses: drinking water supply, fish and shellfish consumption, recreation, aquatic like.

11) groundwater pollutants (nitrate).

12) surface water pollutants.

13) selected coastal surface water pollutants in shellfish.

14) estuarine eutrophication conditions.

15) contaminated sediments.

16) selected point source loadings to surface water and groundwater.

17) nonpoint source sediment loadings from crop land.

18) marine debris.

These environmental indicators cover a wide range of problems and many are for specific local uses. Most, however, are applicable to stormwater problems in urban areas.

Claytor (1996a and 1997) summarized the approach developed by the Center for Watershed Protection as part of their EPA sponsored research on identifying watershed indicators that can be used to assess the effectiveness of stormwater management programs (Claytor and Brown 1996). The indicators selected are direct or indirect measurements of conditions or elements which indicate trends or responses of watershed conditions to stormwater management activities. Categories of these environmental indicators are shown in Table 2, ranging from conventional water quality measurements to citizen surveys. Biological and habitat categories are also represented. Table 3 lists the 26 indicators, by category, while the costs of implementing some of the indicators are shown in Table 5.It is recommended that appropriate indicators be selected from each category for a specific area under study. This will enable a better understanding of the linkage of what is done on the land, how the sources are regulated or managed, and the associated receiving water problems. The indicators were selected to: 1) measure stress or the activities that lead to impacts on receiving waters, 2) assess the resources itself, and 3) measure the regulatory compliance or program initiatives. Claytor (1997) presented a framework for using stormwater indicators, as shown below:

Level 1 (Problem Identification):

1) establish management sphere (who is responsible, other regulatory agencies involved, etc.)

2) gather and review historical data

3) identify local uses which may be impacted by stormwater (flooding/drainage, biological integrity, non-

contact recreation, drinking water supply, contact recreation, and aquaculture).

4) inventory resources and identify constraints (time frame, expertise, funding and labor limitations)

5) assess baseline conditions (use rapid assessment methods).

The selection of the indicators to assess the baseline conditions should be based on the local uses of concern, as shown on Table 6. Most of the anticipated important uses are shown to require indicators selected for each of the categories.

The Level 2 assessment strategy is for examining the local management program and is outlined below:

1) state goals for program (based on baseline conditions, resources, and constraints)

2) inventory prior and on-going efforts (including evaluating the success of on-gong efforts)

3) develop and implement management program

4) develop and implement monitoring program (more quantitative indicators than typically used for the level 1 evaluations above)

5) assess indicator results (does the stormwater indicator monitoring program measure the overall watershed health?)

6) re-evaluate management program (update and revise management program based on measured successes and failures)

Cave (1998) described how environmental indicators are being used to summarize the massive amounts of data being generated by the Rouge River National Wet Weather Demonstration Project in Wayne County (Detroit area), MI. This massive project is examining existing receiving water problems, the performance of stormwater and CSO management practices, and receiving water responses in a 438 mi² watershed having more than 1.5 million people in 48 separate communities. The baseline monitoring program has now more than 4 years of continuous monitoring of flow, pH, temperature, conductivity, and DO, supplemented by automatic sampling for other water quality constituents, at 18 river stations. More than 60 projects are examining the effectiveness of stormwater management practices and 20 projects are examining the effectiveness of CSO controls, each also generating large amounts of data. Toxicants are also being monitored in sediment, water, fish tissue, and with semipermeable membranes to help evaluate human health and aquatic life effects. Habitat surveys were conducted at 83 locations along more than 200 miles of waterway. Algal diversity and benthic macroinvertebrate assessments were also conducted at these survey locations. Electrofishing surveys were conducted at 36 locations along the main river and in tributaries. Several computer models were also used to predict sources, loadings, and wet weather flow management options for the receiving waters and for the drainage systems. A geographic information system was used to manage and provide spatial analyses of the massive amounts of data collected. However, there was still a great need to simply present the data and findings, especially for public presentations. Cave described how they developed a short list of 35 indicators, based on the list of 18 from EPA and with discussions with state and national regulatory personnel. They then developed seven indices that could be color-coded and placed on maps to indicate areas of existing problems and projected conditions based on alternative management scenarios. These indices are described as follows:

Condition Quality Indicators:

1) dissolved oxygen. Concentration and % saturation values (ecologically important)

2) fish consumption index. Based on advisories from the Michigan Dept. of Public Health.

3) river flow. Significant for aquatic habitat and fish communities.

4) bacteria count. E. coli counts based on Michigan Water Quality Standards, distinguished for wet and dry conditions.

Multi-Factor Indices:

1) aquatic biology index. Composite index based on fish and macroinvertebrate community assessments (populations and individuals)

2) aquatic habitat index. Habitat suitability index, based on substrate, cover, channel morphology, riparian/bank condition, and water quality.

3) aesthetic index. Based on water clarity, color, odor, and visible debris.

These seven indicators represent 30 physical, chemical, and biological conditions what directly impact the local receiving water uses (water contact recreation, warmwater fishery, and general aesthetics). Cave presented specific descriptions for each of the indices and gave examples of how they are color-coded for map presentation.

The use of reference sites is common to many bioassessment approaches. As indicated above, reference sites typically are selected as representing as close to natural conditions as possible. However, it is not possible to identify such pristine locations representing varied habitat conditions in most areas of the country. Ohio, for example, has numerous reference sites throughout the state representing a broad range of conditions, but few are completely unimpacted by modifications or human activity in the watersheds. Schueler (1997b) reviewed a USGS report prepared by Crawford and Leant that examined the differences between streams located in forested, agricultural, and urban watersheds in North Carolina. He points out that in many cases, a completely natural forested area is not a suitable benchmark for current conditions before urbanization. In many areas of the country, agricultural land is being converted to urban land, and the in-stream changes expected may be better compared to agricultural conditions. The USGS study found that the stream impacted by agricultural operations was intermediate in quality, with higher nutrient and worse substrate conditions than the urban stream, but better macroinvertebrate and fish conditions. The forested watershed had the best conditions (good quality conditions for all categories), except for somewhat higher sediment heavy metal concentrations than expected. Even though the agricultural watershed had little impervious area, it had high sediment and nutrient discharges, plus some impacted stream corridors. The urban stream had poor macroinvertebrate and fish conditions, poor sediment and temperature conditions, and fair substrate and nutrient conditions.

Almost all states using bioassessment tools have relied on the EPA reference documents as the basis for their programs. Common components of these bioassessment programs (in general order of popularity) include:

· macroinvertebrate surveys (almost all programs, but with varying identification and sampling efforts)

· habitat surveys (almost all programs)

· some simple water quality analyses

· some watershed characterizations

· few fish surveys

· limited sediment quality analyses

· limited stream flow analyses

· hardly any toxicity testing

· hardly any comprehensive water quality analyses

Normally, numerous metrics are used, typically only based on macroinvertebrate survey results, which are then assembled into a composite index. Many researchers have identified correlations between these composite index values and habitat conditions. Water quality analyses in many of these assessments are seldom comprehensive, a possible over-reaction to conventional very costly programs that have typically resulted in minimally worthwhile information. Burton and Pitt (1998) have recommended a more balanced assessment approach, using toxicity testing and carefully selected water and sediment analyses to supplement the needed biological monitoring activities. A multi-component assessment enables a more complete evaluation of causative factors and potential mitigation approaches.

The Milwaukee area may have more stormwater and receiving water data that has been collected over a longer time than any other area. A brief review of some of their notable accomplishments follows, along with a review of a recent study that has built on this historical information to obtain a better and more complete understanding of their local problems and potential stormwater management options.

The personnel of the Wisconsin Department of Natural Resources (WDNR), in conjunction with U.S. Geological Survey and Soil Conservation Service (now Natural Resources Conservation Service) personnel, and personnel from numerous local agencies, have been very active in evaluating local conditions since the early 1970s. They participated in early IJC (International Joint Commission) studies in cooperation with Canada, and the U.S. EPA’s Nationwide Urban Runoff Program (NURP).

The City of Milwaukee has a land area of about 96 mi² and had a 1990 population of about 630,000, although the surrounding urbanized area is comprised of many separate municipalities and covers at least 200 mi² with a population of more than one million. The central part of Milwaukee has a combined sewer system, but the vast majority of the city, plus most of the surrounding communities, have separate storm drainage. The storm drainage flows to Lake Michigan, with most flowing from the Milwaukee, Kinnickinnic, and Menomonee Rivers (which form the inner harbor) from urbanized tributaries (such as Lincoln Creek). The annual rainfall in Milwaukee is about 30 inches, and snow cover is common for several months each winter. Winter temperatures can be severe and snowmelt is an important contributor to urban runoff pollution, in addition to rainfall induced runoff.

Stormwater quality management in the Milwaukee area was initiated as part of the Wisconsin Priority Watershed Program. This program was developed in 1978 to help combat both urban and rural nonpoint sources of pollution. This program is one of the oldest in the nation funding nonpoint pollution abatement. An important element of this program is retrofitting control practices in both rural and urban areas. The program was initially heavily involved in rural areas, with technical assistance from the NRCS. A unique aspect of the program is that it is implemented on a watershed, and not on a political jurisdiction basis. Of the state’s 330 watersheds, 130 (mostly located in the southern part of Wisconsin) will likely require comprehensive management activities to control nonpoint pollutants. A 25 year plan was developed in 1982 which would require the start-up of about eight or nine new watersheds abatement efforts per year. The watershed plans are prepared by the state with cooperation and reviews by local government agencies. They contain detailed analyses of the water resources objectives (existing and desired beneficial uses including the problems and threats to these uses), the critical sources of problem pollutants, and the control practices that can be applied within each watershed. The plans also include implementation schedules and budgets to meet the pollution reduction objectives.

Each plan requires one year to prepare, including the necessary fieldwork. Various field inventory activities are needed to prepare the plans, including aquatic biology and habitat surveys to identify existing and potential fishery uses, streambank surveys to identify the nature and magnitude of streambank erosion problems and to help design needed controls, field and barnyard surveys to supply information needed to estimate and rank their pollution potentials and to design farm control practices, and urban surveys needed to evaluate urban runoff pollution potential and its control.

Urban planning was initiated in 1983 in the Milwaukee and Madison areas, with other urban areas of the state following. The urban practices eligible for cost sharing identified in these plans have included streambank protection, detention basins, and infiltration devices for existing urbanized areas. Construction site erosion controls are also usually required as a condition for a grant agreement in an urban area, but they are not eligible for state cost sharing. About $3 to $5 million per year will be used by the nonpoint source program over a 20 year period in controlling urban runoff (Pitt 1986).

The Wisconsin nonpoint source plan addresses watersheds and not just political areas. The Milwaukee River basin contains 500 stream miles, 100 lakes, and 60,000 acres of wetlands in its 900 mi² watershed. The city of Milwaukee is at the terminus of the river, where it discharges into Lake Michigan. The water quality in the watershed varies dramatically, from excellent in many headwater trout streams to poor in the heavily urbanized southern portions of the basin. The Milwaukee Basin Priority Watershed Program started in 1985 as a voluntary program to address both urban and rural sources of nonpoint pollution. More than 500 rural landowners and 26 local governments have participated in the program, with total local and state investments of about $40 million.

An outcome of the Milwaukee River South Watershed plan included goals for reducing urban stormwater discharges (D’Antuono 1998). These goals were 50% reductions for suspended solids and heavy metals, and 50 to 70% reductions for phosphorus. The city cleanup program has four major elements:

They identified that construction site erosion was the leading cause of sediment into the Milwaukee River South watershed. Of the annual 26,000 tons of sediment, about 62% comes from construction sites, while only 16% comes from cropland, with the remainder from urban stormwater and streambank erosion.

A Wisconsin Pollutant Discharge Elimination System permit (under EPA NPDES authority) was issued to Milwaukee in October 1994. It covers stormwater discharges from more than 200 major outfalls to area streams and Lake Michigan. The permit addresses discharges from Milwaukee and another 26 local communities. These communities each pay the Wisconsin Department of Natural Resources $5,000 per year as a permit fee, totaling more than $1 million for the 5 year permit period.

An important element of the Milwaukee wet weather flow program is monitoring urban streams to identify and quantify actual receiving water problems. Milwaukee participated in the early International Joint Commission studies with Canada to characterize discharges into the Great Lakes. Milwaukee was also a participant of the U.S. EPA’s Nationwide Urban Runoff Program (NURP) in the early 1980s. During NURP, eight single land use catchments were extensively monitored for three years in cooperation with the USGS. The benefits of street cleaning as a pollutant discharge reduction practice were included in this effort. Snowmelt characterization monitoring and effects of de-icing compounds have also been extensively studied in Milwaukee. Recently, detailed studies on toxicant sources, effects, and controls have also been conducted in Milwaukee, including a recent study conducted in the heavily urbanized Lincoln Creek, (having a 19 mi² watershed and being 9 mi long). A seven tiered indicator program, incorporating many physical, chemical, and biological tests were simultaneously conducted which identified long-term toxicity problems, likely associated with resuspended contaminated sediments having high levels of organic compounds. It was found that discharges of these fine sediments could be significantly reduced through the use of well-designed and maintained wet detention basins. The in-stream toxicity monitoring methods developed and used during the Lincoln Creek study can be used by other municipalities to answer the following basic questions:

The benefits of stormwater controls have also been evaluated in Milwaukee, especially grass swales, wet detention ponds, and underground devices for critical source areas. The Southeastern Wisconsin Regional Planning Commission (SEWRPC 1991) also prepared a comprehensive report documenting costs associated with construction site erosion and stormwater control.

The Wisconsin Nonpoint Source Program also includes an important public education component. A survey was conducted in Milwaukee to identify the most likely successful public education program (Simpson 1994). More than 3000 questionnaires were evaluated indicating that TV news stories, newspaper articles, targeted newsletters, and pamphlets would be most effective. Site visits, workshops, and videos were unlikely to be successful. The questionnaire also found that more than 90% of the Milwaukee respondents are willing or are already doing activities to protect water quality (recycling used oil, separating household hazardous wastes, limiting landscaping chemicals, controlling dog wastes, etc.). Virtually all of the respondents rated the local waters as poor to fair and less than 10% used the local waters for any recreational activities. However, more than half were willing to pay more than $50 per household per year for programs to protect and restore local waters.

A notable example of one of the many Milwaukee area studies was conducted by Masterson and Bannerman (1994) in local streams. This project was conducted to document the specific problems occurring in the watershed and to justify the need to continue abatement expenditures. Later projects have since been conducted which have documented sources of problem pollutants and the effectiveness of different control programs. This comprehensive monitoring effort included chemical analyses of outfall water (from the Milwaukee NPDES stormwater permit application efforts), bottom sediments, fish and crayfish tissue, and stream water. In addition, semipermeable polymeric membrane devices (SPMDs) were also placed in numerous stream locations to measure the bioaccumulation potential of PAHs. Biological monitoring examined benthic macroinvertebrates (to supplement other local fish evaluations). Habitat evaluations were also conducted. Side-stream toxicity tests were also conducted during a later study in one of the urbanized streams. Five urbanized streams were examined at several locations in each, along with an upstream reference stream for comparison. Three of the streams were 100% urbanized, and two were about 50% urbanized, while the reference stream was 100% rural. The streams were all 2 to 4^th order streams, and were 6 to 28 miles in length. Overall, the surveys showed that the urban streams were highly degraded and do not support their beneficial use designations (limited aquatic life to warmwater sport fisheries and partial to full body contact recreation).

They found that land use correlated with the amount of biological degradation and limited recreation uses. Pollutant tolerant fish species and macroinvertebrates made up the majority of he organisms found in the urban streams, while the reference stream supported a healthy population of pollutant tolerant and intolerant species. The benthic macroinvertebrate bioassessment scores indicated moderate to severe impairments for the urban streams and nonimpaired conditions for the reference stream. The two creeks that were only 50% urbanized supported a healthier, but still degraded biological community. All of the fecal coliform geometric means in the sampled urbanized Lincoln Creek were 3 to 20 times the criterion for water contact recreation.

The most important potential problem pollutants in the water included the heavy metals (especially copper and zinc which actually exceeded acute toxicity criteria) and the PAHs. Total PAHs, lead, and oil and grease were all several times greater than the values observed in the sediments from the reference stream, with lead (50 to 100 mg/kg) and oil and grease (1,000 to 3,200 mg/kg) frequently exceeding the applicable sediment criteria (60 mg/kg for Pb and 2,000 for oil and grease). None of the whole fish tissue metal concentrations were significantly greater than the fish tissue from the reference stream, but DDT and PAHs were found in urban fish tissue and were not found in the tissue from the reference site. Crayfish lead (0.5 to 2 mg/kg), copper (30 to 40 mg/kg), and PAH (90 to 1100 m g/kg) tissue concentrations were appreciably higher in urban specimens than in reference specimens (<0.1 Pb mg/kg, 18 Cu mg/kg, and <5 PAH m g/kg), however. The SPMD PAH values were also very high in the urban streams (9,000 to 50,000 m g/kg) compared to the values found in the rural reference stream (700 m g/kg). Stream flows were also much flashier and greater in the urbanized streams compared to nonurbanized streams for similar storm conditions (see Figure 4).

Figure 4. Hydrographs for urbanized Lincoln Creek and nonurbanized Jackson Creek (Masterson and Bannerman 1994).

They concluded that biological integrity of urban streams is dependent on the inter-relationship of many elements, as indicated on Figure 5. Stormwater discharges affect physical characteristics of high and low flows, plus add to sediment accumulations which severely degrade most urban stream habitats. Discharged pollutants also cause exceedences of some water quality criteria (especially heavy metals, PAHs, and bacteria), plus contaminate stream sediments, and possibly lead to bioaccumulation of toxicants in fish, and possibly even cause toxic conditions (mostly chronic toxicity).

Figure 5. Relationship between stormwater discharges and the biological integrity of urban streams (Masterson and Bannerman 1994).

There are several mechanisms where stormwater exposure can cause potential human health problems. These include exposure to stormwater contaminants at swimming areas affected by stormwater discharges, drinking water supplies contaminated by stormwater discharges, and the consumption of fish and shellfish that have been contaminated by stormwater pollutants. Understanding the risks associated with these exposure mechanisms is difficult and not very clear. Receiving waters where human uses are evident are usually very large and the receiving waters are affected by many sanitary sewage and industrial point discharges, along with upstream agricultural nonpoint discharges, in addition to the local stormwater discharges. In receiving waters only having stormwater discharges, it is well known that inappropriate sanitary and other wastewaters are also discharging through the storm drainage system. These "interferences" make it especially difficult to identify specific cause and effect relationships associated with stormwater discharges alone, in contrast to the many receiving water studies that have investigated ecological problems, noted previously, that can more easily study streams affected by stormwater alone. Therefore, much of the human risk assessment associated with stormwater exposure must use theoretical evaluations relying on stormwater characteristics and laboratory studies in lieu of actual population studies. However, some site investigations, especially related to swimming beach problems associated with nearby stormwater discharges, have been conducted. This section presents a summary of the human health effects of stormwater discharges, stressing these swimming beach studies.

As noted previously in Chapter 1, there are several impediments associated with the reuse of stormwater in residential areas. The most serious problems appear to be associated with the presence of potential pathogens in problematic numbers. Contact recreation in pathogen contaminated waters has been studied at many locations. The sources of the pathogens is typically assumed to be sanitary sewage effluent, or periodic industrial discharges from certain food preparation industries (especially meat packing and fish and shellfish processing). However, several studies have investigated pathogen problems associated with stormwater discharges. It has generally been assumed that the source of the pathogens in the stormwater are from inappropriate sanitary connections. However, as will be shown in Chapters 5 and 6, stormwater unaffected by these inappropriate sources still contains high counts of pathogens that are also found in surface runoff samples from many urban surfaces. Needless-to-say, sewage contamination of urban streams is an important issue that needs attention during an urban water resource investigation. Therefore, the following subsection presents a brief overview of this contamination.

The following case studies present summaries of various studies conducted throughout the U.S. that investigated contamination of urban streams that were only supposed to be receiving stormwater discharges. Many of the problematic discharges were from sanitary sewage. Obviously, inappropriate discharges must be identified and corrected as part of any effort to clean up urban streams. If these sources are assumed to be non-existent in an area and are therefore not considered in the stormwater management activities, incorrect and inefficient management decisions are likely, with disappointing improvements in the receiving waters. Chapter 9 presents a strategy that was developed by Lalor (1993), Pitt, et al. (1993), and Pitt and Lalor (1997) for the EPA to support the outfall screening activities required by the NPDES Stormwater Permit Program (described in Chapter 15) to identify and correct inappropriate discharges to storm drainage systems. Chapter 14 also describes CSO and SSO management strategies that can be used for correcting sanitary sewage discharges to urban streams.

A number of issues emerged from the individual projects of the U.S. EPA’s Nationwide Urban Runoff Program (NURP) (EPA 1983). One of these issues involved illicit connections to storm drainage systems and was summarized as follows in the Final Report of the NURP executive summary: "A number of the NURP projects identified what appeared to be illicit connections of sanitary discharges to stormwater sewer systems, resulting in high bacterial counts and dangers to public health. The costs and complications of locating and eliminating such connections may pose a substantial problem in urban areas, but the opportunities for dramatic improvement in the quality of urban stormwater discharges certainly exist where this can be accomplished. Although not emphasized in the NURP effort, other than to assure that the selected monitoring sites were free from sanitary sewage contamination, this BMP (Best Management Practice) is clearly a desirable one to pursue." The illicit discharges noted during NURP were especially surprising, because the monitored watersheds were carefully selected to minimize factors other than stormwater. Presumably, illicit discharge problems in typical watersheds would be much worse. Illicit entries into urban storm sewerage were identified by flow from storm sewer outfalls following substantial dry periods. Such flow could be the result of direct "illicit connections" as mentioned in the NURP final report, or could result from indirect connections (such as contributions from leaky sanitary sewerage through infiltration to the separate storm drainage). Many of these dry-weather flows are continuous and would therefore also occur during rain induced runoff periods. Pollutant contributions from the dry-weather flows in some storm drains have been shown to be high enough to significantly degrade water quality because of their substantial contributions to the annual mass pollutant loadings to receiving waters.

Washtenaw County (Ann Arbor), MI

From 1984 to 1986, Washtenaw County, Michigan, dye-tested 160 businesses in an effort to locate direct illicit connections to the county stormwater sewerage (Murray 1985; Schmidt and Spencer 1986; Washtenaw County 1988). Of the businesses tested, 61 (38%) were found to have improper storm drain connections. The Huron River Pollution Abatement Program was the most thorough investigation of such improper connections. Beginning in 1987, 1067 businesses, homes and other buildings located in the Huron River watershed were dye tested. The following results were reported. Illicit connections were detected at 60% of the automobile related businesses inspected, including service stations, automobile dealerships, car washes, and auto body and repair shops. All plating shops inspected were found to have improper storm drain connections. Additionally, 67% of the manufacturers tested, 20% of the private service agencies, and 88% of the wholesale/retail establishments tested were found to have improper storm sewer connections. Of 319 homes dye tested, 19 were found to have direct sanitary connections to storm drains. The direct discharge of rug cleaning wastes into storm drains by carpet cleaners was also noted as a common problem. Several surveys, beginning as early as 1963, identified bacterial and chemical contamination of the Allen Creek storm drainage system. Studies in 1963, 1978 and 1979 found that discharges from the Allen Creek storm drain contained significant quantities of fecal coliform and fecal streptococci. The 1979 study also documented high pollutant loads of solids, nitrates and metals. A large number of inappropriate storm drain connections originating from businesses were found, especially within automobile related facilities. Chemical pollutants, such as detergents, oil, grease, radiator wastes and solvents were causing potential problems.

The elimination of these storm drain connections prevented thousands of gallons of contaminated water from entering the Huron River from the Allen Creek storm drainage system annually. Eight sampling locations along the main stem and major lateral branches of the storm drainage system were established and monitored for 37 chemicals during rain events. From 1984 to 1986, 32 (86%) of these chemicals showed a decrease in concentrations while only 2 (5%) showed an increase. In spite of this improvement, chemical concentrations in the stormwater discharges at the Allen Creek outfall were still greater than those from the control station much of the time.

This program has been underway since June of 1985 (Falkenbury 1987). Investigations to date indicate few direct connections from industries to storm drains. Storm runoff, in addition to illegal dumping, accidental spills and direct discharges into the street or adjacent creeks seem to account for the majority of the contaminants entering the storm drainage system. Major problems stemmed from septic tanks, self-management of liquid wastes by industry and construction of municipal overflow bypasses from the sanitary sewer to the storm drains. The success of this program was judged by a decline in the number of undesirable features at the target outfalls. An average of 44 undesirable observations per month were made in 1986 (522 total), compared to an average of 21 undesirable observations per month in 1988.

In 1987, an inspection of the 90 urban stormwater outfalls draining into Inner Grays Harbor in Washington revealed 29 (32%) flowing during dry weather (Beyer, et al. 1979; Pelletier and Determan 1988). A total of 19 outfalls (21%) were described as suspect, based on visual observation and/or anomalous pollutant levels, as compared to those expected in typical urban stormwater runoff characterized by NURP. At least one storm drain system was later found to receive a residential sanitary sewage connection which has since been corrected. This drain exhibited no unusual visual characteristics, but was found to have atypical pH and total suspended solids levels. Notably, fecal coliform levels were within the typical range expected for stormwater.

A Sacramento, California, investigation of urban discharges identified commercial as well as domestic discharges of oil and other automobile related fluids as a common problem based on visual observations (Montoya 1987). Montoya found that slightly less than half the water discharged from Sacramento's stormwater drainage system was not directly attributable to precipitation. Most of this water comes from unpermitted sources, including illicit and/or inappropriate entries to the storm drainage system.

During the Bellevue, Washington Urban Runoff Project baseflows as well as stormwater from two residential urban basins were monitored (Pitt 1985; Pitt and Bissonnette 1984). The areas included in this study, Surrey Downs and Lake Hills, are about 5 km apart and each covered an area of about 40 ha. Both were fully developed, with predominantly single family residences. No septic tanks were present in either area and the storm drainage systems were thoroughly mapped and investigated to ensure no non-stormwater discharges to storm drainage systems or obvious illegal discharges. The Bellevue, Washington, NURP project also summarized the reported incidents of intermittent discharges and dumpings of pollutants into the local storm drainage system. During a three year period of time, about 50 citizen contacts were made to the Bellevue Storm and Surface Water Utility District concerning water quality problems. About 25 percent of the complaints concerned oil being discharged into storm drain inlets. Another important category of complaints was for aesthetic problems, such as turbid or colored water in the creeks. Various industrial and commercial discharges into the storm drainage system were detected. Concrete wastes flushed from concrete trucks at urban job sites were a frequently occurring problem. Cleaning establishment discharges into creeks were also a common problem. Vehicle accidents also resulted in discharges of gasoline, diesel fuel, hydraulic fluids, and lawn care chemicals into the storm drain inlets.

A field screening program was conducted to determine the relative levels of contamination at various locations in the Stony Brook drainage system (Metcalf and Eddy 1994). During eight days of dry-weather sampling, numerous inappropriate discharges of sanitary sewage into the drainage system were identified using the investigative procedures developed by Pitt, et al. (1993) and a modified flow chart approach.

Water Environment & Technology (1996a) reported that the fecal coliform counts decreased from about 500 counts/100 mL to about 150 counts/100 mL in the Mississippi River after the sewer separation program in the Minneapolis and St. Paul area of Minnesota. Combined sewers in 8,500 ha were separated during this 10-year, $332 million program.

The Executive Committee of the Metropolitan Toronto Council has a Water Pollution Committee that aggressively examined different responses to the fecal contamination found on Toronto’s beaches along the city’s waterfront in 1983. Lees (1984) presented a comprehensive review of the political responses and interpretations of the data. It was estimated that about $250 million, to be spent over ten years, would be needed as part of a beach cleanup effort. Much of that cost would have been spent to lengthen the discharge pipes from the treatment plant further out into Lake Ontario. This large cost estimate led to a number of studies that examined the likely causes of the high fecal coliform counts found in the swimming beach areas. In addition, the methods of sample collection were also reviewed.

The city’s board of health had been gathering water samples from swimming beaches since 1956. Initially, the samples were collected by manually scooping water into sample bottles from a harbor police power boat, in at least several feet of water. As part of the Toronto Area Watershed Management Strategy Study (TAWMSS), swimming beach water samples were collected from shore, from very shallow water. These near-shore samples showed very high fecal coliform (FC) counts (to more than 100,000 FC/100 mL), much greater than the Ontario standard of 100 FC/100 mL. There was great concern in the apparent degradation of the beach water, but it was realized that the differences in the sampling locations was likely responsible for the large differences in the observed contamination levels. However, the shallow water levels were of concern due to exposure to many bathers, especially children.

Additional local studies were therefore conducted to investigate the reasons for these high counts, with typical conclusions being sea gulls, Canada geese, storm sewers, pet wastes, CSOs, etc. In one of these studies, Pitt and McLean (1986) and GLA (1983) monitored and characterized both stormwater and baseflows in test catchments. The project involved intensive monitoring in two test areas. The Emery catchment area, located near the City of North York, covered approximately 154 ha with predominantly "medium" industrial land uses (processing goods for final consumption). The Thistledown catchment, located in the City of Etobicoke, covered approximately 39 ha with residential and commercial land uses. During cold weather, the increases in dissolved solids were quite apparent in baseflows and snowmelt for both study catchments. This increase was probably caused by high chlorides from road salt applications. In contrast, bacteria populations were noticeably lower in all outfall discharges during cold weather. Nutrient and heavy metal concentrations at the outfalls remained fairly constant during cold and warm weather. Either warm- or cold-weather baseflows were responsible for most of the yields for many constituents from the industrial catchment. These constituents included runoff volume, phosphorus, total Kjeldahl nitrogen, chemical oxygen demand and chromium. Important constituents that had high yields in the baseflow from the residential/commercial catchment included total solids, dissolved solids, chlorides, and fecal coliform and Pseudomonas aeruginosa bacteria.

Gartner Lee and Associates, Ltd. (GLA 1983) conducted an extensive survey of dry-weather flows in storm drainage systems in the Humber River watershed (Toronto) in an attempt to identify the most significant urban runoff pollutant sources. About 625 outfalls were sampled two times during dry-weather, with analyses conducted for many pollutants, including organics, solids, nutrients, metals, phenols, and bacteria. About 59% had dry-weather flows, and about 33% of the outfalls were discharging at rates greater than 1 L/sec. The dry-weather flows were found to contribute significant loadings of nutrients, phenols, and metals, compared to upstream conditions. About 10 to 14 percent of the outfalls were considered significant pollutant sources. Further investigations identified many industrial and sanitary sewage non-stormwater discharges into the storm drainage. An apartment building with the sanitary drains from eight units illegally connected to the storm drainage system was typical of the problems found. Other problem areas were found in industrial areas, including yard storage of animal hides and yard runoff from meat packing plants.

Visual inspection of stormwater pipes discharging to the Rideau River (Ontario) found leakage from sanitary sewer joints or broken pipes to be a major source of storm drain contamination (OME 1983). A study of the lower Rideau River in the Regional Municipality of Ottawa-Carleton was conducted to establish the causes of bacteriological water quality degradation in the urbanized reach of the river and to analyze the impacts of future urbanization. Earlier programs had identified and corrected many cross-connections between sanitary sewers and stormwater sewers. Bacteriological water quality improved, but swimming standards at beaches were still not obtained.

During the development of the methods to investigate inappropriate discharges, a three-mile section of Village Creek in Birmingham, AL, was selected for field verification of the test methods (Pitt, et al. 1993, Pitt and Lalor 1997). The drainage area for this section of the creek contains about 4500 acres. Residential land use comprises approximately 88% of the total area, commercial land use approximately 8%, and industrial land use less than 1%. The majority of the drainage area is serviced by sanitary sewers, but some septic tanks are also used. A total of 65 stormwater outfalls were located. Outfall diameters ranged from 2 inches to 12 feet, excluding open ditches. All sites were visited at least 8 times during the field investigation period. Of these 65 outfalls, 48 (74%) were always dry, 6 (9%) had flow intermittently, and 11 (17%) were always flowing. Eighteen direct unpermitted discharges to the creek from nearby industries and commercial areas were also located; 10 (56%) were always dry, 6 (33%) had intermittent flow, and 2 (11%) were always flowing. The dry weather flows from two of the 65 outfalls were found to be mostly sanitary sewage, while the flows from another nine were predominately washwaters. The remaining outfalls with dry weather flows were mostly affected by natural waters (most likely groundwater infiltration) or leaking domestic water.

Periodic stream surveys of tributaries of the Cahaba River in the Birmingham area (mostly the Little Cahaba River, upstream of Lake Purdy) during summer months have found that the small river contained about 1/3 treated sewage from upstream poorly operated municipal treatment facilities (since corrected), septage from failing septic tanks, and SSO discharges.

During a current EPA sponsored project investigating SSO discharges being conducted by Lalor at UAB, sewage, through SSOs and poorly operating septic tanks, were found to make up about 25% of the dry weather flows in the small, completely urbanized stream in Homewood, AL, being studied. However, sewage contributions in the much larger, and much less urbanized 5-mile Creek are very small (on a percentage standpoint), although SSOs exist in the urbanized area. These streams are still being evaluating, including future human health risk assessments associated with these discharges.

Therefore, urban stormwater runoff includes waters from many other sources which find their way into storm drainage systems, besides from precipitation. There are cases where pollutant levels in storm drainage are much higher than they would otherwise be because of excessive amounts of contaminants that are introduced into the storm drainage system by various non-stormwater discharges. Additionally, baseflows (during dry weather) are also common in storm drainage systems. Dry-weather flows and wet-weather flows have been monitored during numerous urban runoff studies. These studies have found that discharges observed at outfalls during dry weather were significantly different from wet-weather discharges and may account for the majority of the annual discharges for some pollutants of concern from the storm drainage system.

In many cases, sanitary sewage was an important component (although not necessarily the only component) of the dry weather discharges from the storm drainage systems. From a human health perspective (associated with pathogens), it may not require much raw or poorly treated sewage to cause a receiving water problem. However, at low discharge rates, the DO receiving water levels may be minimally affected. The effects these discharges have on the receiving waters is therefore highly dependent on many site specific factors, including frequency and quantity of sewage discharges and the creek flows. In many urban areas, the receiving waters are small creeks in completely developed watersheds. These creeks are the most at risk from these discharges as dry base flows may be predominately dry weather flows from the drainage systems. In Tokyo (Fujita 1998), for example, numerous instances were found where correcting inappropriate sanitary sewage discharges resulted in the urban streams losing all of their flow. In cities that are adjacent to large receiving waters, these discharges likely have little impact (such as DO impacts from Nashville CSO discharges on the Cumberland River, Cardozo, et al. 1994). The presence of pathogens from raw, or poorly treated sewage, in urban streams, however, obviously presents a potentially serious public health threat. Even if the receiving waters are not designated as water contact recreation, children are often seen playing in small city streams.

Epidemiology can be defined as the study of the occurrence and causes of disease in human populations and the application of this knowledge to the prevention and control of health problems. The general population often views epidemiology and associated risk assessments with skepticism when risks associated with seemingly everyday activities are quantified, especially when associated with periodic "food scares" that are typically exaggerated or misinterpreted in the press. Technical experts also may feel uncomfortable with the results of epidemiological studies because of the typically very low numbers of affected people in a study population. However, much of the information that is used in developing environmental regulations protecting human health originates with epidemiological studies and a more through understanding of the science of epidemiology would dispel much of the confusion associated with these studies.

Epidemiology has routinely been used to assess risks associated with contaminants in drinking waters. Epidemiology has also recently been used to investigate human health risks associated with swimming in waters contaminated by stormwater. However, Craun, et al. (1996) state that the results of environmental epidemiology studies (the assessment of human health effects associated with environmental contaminants, where indicators of disease are mostly studied instead of the disease itself) have provoked controversy. Their excellent review article on epidemiology applied to water and public health discusses many of these problems and offers suggestions to enable better interpretation of existing studies and better design of future studies. The following paragraphs are summarized from their article.

The definition of terms is important. For example, epidemiologists use several measures to describe disease frequency. Incidence is the rate at which new cases of disease occur, whereas prevalence measures both new and existing cases in the total population. Therefore, prevalence is "the proportion of people who have a specific condition at any specific time" and is typically measured as a percentage of the total population. Incidence considers the duration of exposure, and the incidence rate may be expressed as the number of cases observed per person-years of exposure, for example. The attack rate is a measure of the cumulative incidence during an outbreak of the disease, and is usually expressed in terms of numbers of cases of disease per population unit (such as per 10,000 people in the population). Secondary outbreaks can also occur for communicable disease and the secondary attack rate refers to the cases of disease attributed to exposure to people having the disease during the primary attack. The secondary rate is usually expressed in terms of susceptible contacts. Geographic-specific (such as part of town receiving water from a specific source) and vehicle-specific (such as waterborne specific disease) attack rates help to determine the source of the disease. Attack rates can also be examined in terms of water consumption by separating the attack rate into different categories associated with different amounts of water consumed, for example. Mortality rate and case fatality rate are also measures of disease frequency. The mortality rate indicates the number of deaths from a certain disease, and or time period, per the total population. The case fatality rate is the proportion of diagnosed individuals having the disease who die of the disease. The crude rates should be standardized to account for differences in demographic characteristics of the population, especially age.

Association is a measure of the dependence between exposure to a contaminant and the onset of disease, but does not necessarily indicate a cause and effect relationship between the variables. Both experimental (clinical or population) and observational (descriptive or analytical) epidemiologic studies are used to determine associations. In clinical experimental studies, active intervention may be used to expose the subjects to specific doses of an infective agent to determine the infective dose of a pathogen, for example. In population experimental studies, the population may be randomly grouped according to different levels of drinking water treatment, and the households would then be extensively examined to determine any differences in disease outbreaks. In descriptive observational studies, information is available about the occurrence of disease and about exposure to specific compounds, exposure periods, and different demographic information. Analytical observational studies test specific hypotheses to evaluate associations between exposure and disease and to confirm the mode of transmission. Ecological studies (or correlation or aggregate studies) examine associations between routinely gathered health and demographic statistics and available environmental measures (such as drinking water constituent concentrations). These studies are typically controversial because the statistical demographic information pertains to groups (lumped information which makes it difficult to identify confounding factors or to normalize) and not to individuals within the groups. Difficulties also relate to incomplete information concerning potential causative agents. Therefore, analytical observational studies (where individuals are studied and more detailed information concerning the potential causative agents can be obtained) should be used to follow up hypotheses developed in ecological studies.

The experimental design of epidemiological studies is very critical. The study must be of sufficient size and have adequate statistical power to detect the hypothesized association. Randomness is also very critical in epidemiological studies to control systematic errors. In most cases, epidemiological studies compare disease rates between a test and a control population. Positive associations (where there is a statistically significant difference between the rates of the two groups) can be caused by random errors. This likelihood can be estimated by calculating the confidence interval of the statistical significance of the association. However, statistical significance (even at a very high level) does not imply a cause and effect relationship between the hypothesized factor and disease. Statistical power can be used to identify the minimum risk that a study is capable of detecting. An environmental epidemiological study should not be conducted "unless the exposure assessment is expected to be reasonably appropriate and accurate." Adequate and complete data to make the exposure assessment must be assured before the study is conducted.

Interpreting associations is based on examining the rate differences (RD), which is the absolute differences in the two rates (incidence rate of disease for the test, or exposed, group minus the incidence rate of disease for the control, or unexposed, group), or the rate ratio (RR), which is the ratio of the rates from the two groups. The odds ratio (OR) is the ratio of the odds of disease of the test group to the odds of disease of the control group, and is interpreted similarly to the rate ratio. If the RR or OR is close to 1.0, there is no association or increased risk between the two groups. If the ratio is 1.8, there is an 80 percent increased risk of disease for the exposed individuals, compared to the unexposed group. The confidence interval of the ratio is used to identify significance of the association. A 95 percent confidence interval of 1.6 to 2.0 signifies a statistically significant estimate because the range does not include 1.0. The relatively narrow range also implies a precise estimate of the association. In contrast, a 95 percent confidence interval of 0.8 to 14.5 does not signify a significant difference because the range includes the value of 1.0. In addition, the wide range also implies an imprecise estimate of the association. Craun, et al. (1996) presents Table XX1 (from Monson 1980) indicating different rate ratios and strengths of associations. Weak associations (ratios of <1.5) are difficult to interpret. Very large range ratios are unlikely to be completely explained by unidentified or uncontrolled confounding characteristics. However, the magnitude of the rate ratio has no bearing on the likelihood that the association is attributed to bias, but causal association cannot be ruled out simply because of a weak association. In many environmental epidemiological studies, the rate ratio is frequently smaller than 1.5, causing speculation that the association may actually be caused by bias. "High quality exposure and study design are important for interpreting risks of this magnitude."

With the low rate ratios frequently encountered in environmental epidemiological studies, cautious interpretations are necessary. Craun, et al. (1996) present the following criteria that are used to assess associations and causality:

Therefore, an effective and convincing interpretation can be supported if many of these above factors are successfully addressed by an environmental epidemiological study.

There have been a few epidemiology studies recently published describing the increased health risks associated with contaminated dry weather flows affecting public swimming beaches. The following discussion presents an overview of the development of water quality criteria for water contact recreation, plus the results of several epidemiological studies that have specifically examined human health problems associated with swimming in contaminated water, including water affected by stormwater. In most cases, the levels of indicator organisms and pathogens causing increased illness were well within the range found in urban streams.

Human health standards for body contact recreation (and for fish and water consumption) are based on indicator organism monitoring. Monitoring for the actual pathogens, with few exceptions, requires an extended laboratory effort, is very costly and not very accurate. Therefore, the use of indicator organisms has become established. Dufour (1984a) presents an excellent overview of the history of indicator bacterial standards and water contact recreation, summarized here. Total coliforms were initially used as indicators for monitoring outdoor bathing waters, based on a classification scheme presented by W.J. Scott in 1934. Total coliform bacteria refers to a number of bacteria including Escherichia, Klebsiella, Citrobacter, and Enterobacter (DHS 1997). They are able to grow at 35^oC and ferment lactose. They are all gram negative asporogenous rods and have been associated with feces of warm blooded animals. They are also present in soil. Scott had proposed four classes of water, with total coliform upper limits of 50, 500, 1,000, and >1,000 MPN/100 mL for each class. He had developed this classification based on an extensive survey of the Connecticut shoreline where he found that about 93% of the samples contained less than 1,000 total coliforms per 100 mL. A sanitary survey classification also showed that only about 7% of the shoreline was designated as poor. He therefore concluded that total coliform counts of <1,000 MPN/100 mL probably indicated acceptable waters for swimming. This standard was based on the principle of attainment, where very little control or intervention would be required to meet this standard. In 1943, the state of California independently adopted an arbitrary total coliform standard of 10 MPN/1 mL (which is the same as 1,000 MPN/100 mL) for swimming areas. This California standard was not based on any evidence, but it was assumed to relate well with the drinking water standard at the time.

Dufour points out that a third method used to develop a standard for bathing water quality used an analytical approach adopted by H.W. Streeter in 1951. He used a ratio between Salmonella and total coliforms, the number of bathers exposed, the approximate volume of water ingested by bathers daily, and the average total coliform density. Streeter concluded that water containing <1,000 MPN total coliforms/100 mL would pose no great Salmonella typhosa health hazard. Dufour points out that it is interesting that all three approaches in developing a swimming water criterion resulted in the same numeric limit.

One of the earliest bathing beach studies to measure actual human health risks associated with swimming in contaminated water was directed by Stevenson (1953), of the U.S. Public Health Service’s Environmental Health Center, in Cincinnati, Ohio, and was conducted in the late 1940s. They studied swimming at Lake Michigan at Chicago (91 and 190 MPN/100 mL median total coliform densities), the Ohio River at Dayton, KY (2,700 MPN/100 mL), at Long Island Sound at New Rochelle and at Mamaroneck, NY (610 and 253 MPN/100 mL). They also studied a swimming pool in Dayton, KY. Two bathing areas were studied in each area, one with historically poorer water quality than the other. Individual home visits were made to participating families in each area to explain the research program and to review the calendar record form. Follow up visits were made to each participating household to insure completion of the forms. Total coliform densities were monitored at each bathing area during the study. More than 20,000 persons participate in the study in the three areas. Almost a million person-days of useable records were obtained. The percentage of the total person-days when swimming occurred ranged from about 5 to 10 percent. The number of illnesses of all types recorded per 1,000 person-days varied from 5.3 to 8.8. They found an appreciably higher illness incidence rate for the swimming group, compared to the nonswimming group, regardless of the bathing water quality (based on total coliform densities). A significant increase in gastrointestinal illness was observed among the swimmers who used one of the Chicago beaches on three days when the average coliform count was 2,300 MPN/100 mL. The second instance of positive correlation was observed in the Ohio River study where swimmers exposed to the median total coliform density of 2,700 MPN/100 mL had a significant increase in gastrointestinal illness, although the illness rate was relatively low. They suggested that the strictest bacterial quality requirements that existed then (as indicated above, based on Scott’s 1934 work) might be relaxed without significant detrimental effect on the health of bathers.

It is interesting to note that in 1959, the Committee on Bathing Beach Contamination of the Public Health Laboratory Service of the UK concluded that "bathing in sewage-polluted seawater carries only a negligible risk to health, even on beaches that are aesthetically very unsatisfactory" (Cheung, et al. 1990 and Alexander, et al. 1992).

Dufour (1984a) pointed out that total coliforms were an integral element in establishing fecal coliform limits as an indicator for protecting swimming uses. Fecal coliform bacteria are a subgroup of the total coliform group. They grow at 44.5^oC and also ferment lactose. They are restricted to the feces of warm blooded animals and can be used to separate bacteria of soil and animal origin (DHS 1997). They do survive for variable periods of time in fecal contaminated soil and water, however. As a result of the Stevenson (1953) study, reported above, a geometric mean fecal coliform level of 200 MPN per 100 mL was recommended by the National Technical Advisory Committee (NTAC) of the Federal Water Pollution Control Administration in 1968 and was adopted by the U.S. Environmental Protection Agency in 1976 as a criterion for direct water contact recreation (Cabelli, et al. 1979). This criterion was adopted by almost all states by 1984. It was felt that fecal coliforms was more specific to sewage contamination and had less seasonal variation that total coliforms. Since fecal coliform exposures at swimming beaches had never been linked to disease, the NTAC reviewed the USPHS studies, as published by Stevenson (1953). The 2,300 MPN/100 mL total coliform count association with gastrointestinal disease was used in conjunction with a measured ratio of fecal coliform to total coliform counts (18%) obtained at the Ohio River site studied earlier. It was therefore assumed that a health effect could be detected when the fecal coliform count was 400 MPN/100 mL (18% of 2,300 = 414). Dufour (1984a) pointed out that a detectable health effect was undesirable and that the NTAC therefore recommended a limit of 200 MPN/100 mL for fecal coliforms. Dufour (1984a) points out that, although likely coincidental, the 1968 proposed limit for fecal coliforms (200 MPN/100 mL) was very close to being theoretically equivalent to the total coliform limit of 1,000 MPN/100 mL that was being replaced (200/0.18 = 1100).

Dufour (1984a) lists the ideal characteristics of bacterial indicators of fecal contamination, as presented by various authors. The authors were in agreement concerning many of the criteria (correlation to pathogens, unable to grow in aquatic environments, more resistant to disinfection than pathogens, and easy to isolate and enumerate), but two important aspects were seldom mentioned, namely that the indicator should have a direct relationship to fecal contamination, and that the indicator density should correlate with health hazards. Many of the follow-up studies conducted since the mid 1970s examined these additional criteria.

E. coli, a member of the fecal coliform group, has been used as a better indicator of fresh fecal contamination. Table XX5 indicates the species and subspecies of the Streptococcus and Enterococcus groups of bacteria that are used as indicators of fecal contamination (DHS 1997).

Source: DHS (1997)

Fecal streptococci bacteria are indicators of fecal contamination. The enterococcus group is a subgroup that is considered a better indication of human fecal contamination. S. bovis and S. equinus are considered related to feces from non-human warm blooded animals (such as from meat processing facilities, dairy wastes, and feedlot and other agricultural runoff), indicating that enterococcus may be a better indication of human feces contamination. However, S. facealis subsp. liquifaciens is also associated with vegetation, insects, and some soils (DHS 1997).

The Cabelli, et al. (1979) study was undertaken to address many remaining questions pertaining to bathing in contaminated waters. Their study examined conditions in New York (at a Coney Island beach, designated as barely acceptable, and at a Rockaway beach, designated as relatively unpolluted). About 8,000 people participated in the study, approximately evenly divided between swimmers and nonswimmers at the two beaches. Total and fecal coliforms, Escherichia, Klebsiella, Citrobacter-Enterobacter, Enterococci, Pseudomonas aeruginosa, and Clostridium perfringens were evaluated in water samples obtained from the beaches during the epidemiological study. The most striking findings were the increases in the rates of vomiting, diarrhea, and stomachache among swimmers relative to nonswimmers at the barely acceptable beach, but not at the relatively unpolluted beach. Ear, eye, nose, and skin symptoms, as well as fever, were higher among swimmers compared to nonswimmers at both beaches. They concluded that measurable health effects do occur at swimming beaches that meet the existing health standards. Children, Hispanic Americans, and low-middle socioeconomic groups were identified as the most susceptible portions of the population.

Cabelli, et al. (1982) presented data from the complete EPA sponsored swimming beach study, conducted in New York, New Orleans, and Boston. The study was conducted to address issues from prior studies conducted in the 1950s (including Stevenson’s 1953 study noted above) that were apparently contradictory. They observed a direct, linear relationship between highly credible gastrointestinal illness and enterococci. The frequency of gastrointestinal symptoms also had a high degree of association with distance from known sources of municipal wastewater. Table XX4 shows correlation coefficients for total gastrointestinal (GI) and highly credible gastrointestinal (HCGI) symptoms and mean indicator densities found at the New York beaches from 1970 to 1976. The best correlation coefficients were found for enterococci. In contrast, the correlation coefficients for fecal coliforms (the basis for most federal and state guidelines) were poor. Very low levels of enterococcus and Escherichia coli in the water (about 10 MPN/100 mL) were associated with appreciable attack rates (about 10/10,000 persons).

Figure XX1 shows regressions of swimming associated gastrointestinal symptom rates (swimmer rates minus nonswimmer rates) against the mean enterococcus and E. coli densities of the water samples. The results clearly show that the risk of gastrointestinal symptoms associated with swimming in marine waters contaminated with municipal wastewater is related to the quality of the water, as indicated by the enterococcus density of the water. They also felt there was a strong case for causality between enterococci and gastrointestinal symptoms, based on the good association, the consistency at the different locations over different years, the reasonable nature of the relationship between enteric disease and fecal contamination, and the coherent association based on observations of waterborne disease transmission during prior outbreaks.

Figure XX1. Regressions of Gastrointestinal Symptom Rates (per 1,000 swimmers) against Enterococcus and E. coli. Densities at Marine Swimming Beaches (Cabelli, et al. 1982).

They concluded that swimming in even marginally polluted marine bathing water is a significant route of transmission for observed gastrointestinal illness. They felt that the gastrointestinal illness was likely associated with the Norwalk-like virus that had been confirmed in 2,000 cases at a shellfish associated outbreak in Australia and at several outbreaks associated with contaminated drinking water.

Fleisher (1991) reevaluated this marine swimming beach data and concluded that the limitation for enterococci promulgated by the EPA in 1986, based on the Cabelli, et al. (1982) study, (35 per 100 mL, geometric mean for 5 equally spaced samples over a 30-day period, for both fresh and saline water) was too severe, due to minor adjustments of the observed data. He was also especially concerned with the use of a single criterion based on pooled data, while the data from the individual sites indicated very different probabilities of gastroenteritis among swimmers at Boston compared to New York and Lake Pontchartrain (which were similar). He also reported that previous studies found bacteria indicator, and possibly pathogen, survival to be inversely correlated with salinity. He therefore concluded that any relation between enterococci and disease causing pathogens may be site specific, possibly related to water salinity. This EPA enterococci criterion for swimming waters was based on an "acceptable" rate of gastroenteritis of 19 cases per 1,000 swimmers, the same rate upon which the fecal coliform criterion (200 MPN/100 mL) was based. It is interesting to note that Fleisher later participated in additional epidemiological studies in the UK and concluded that 33 fecal streptococci (essentially enterococci)/100 mL was the threshold of increased risk for gastrointestinal illness for swimmers (Kay, et al. 1994).

Dufour (1984a) also reviewed a series of studies conducted at freshwater swimming beaches from 1979 to 1982, at Tulsa, OK, and at Erie, PA. Only enterococci, E. coli, and fecal coliforms were monitored, based on the results of the earlier studies. Table XX111 shows the correlation coefficients for these three bacterial parameters and gastrointestinal disease.

Table XX111. Correlation Coefficients for Bacterial Parameters and Gastrointestinal Disease

(Fresh Water Swimming Beaches)

These results are quite different than the results from the marine studies, in that both enterococci and E. coli had high correlation coefficients between the bacterial levels and the incidence of gastrointestinal illness. However, the result was the same for fecal coliforms, in that there was no association between fecal coliform levels and gastrointestinal illness. Dufour (1984b) concluded that enterococci would be the indicator of choice for gastrointestinal illness, based on scientific dependability. E. coli could also be used, if only fresh waters were being evaluated. Fecal coliforms would be a poor choice for monitoring the safety of bathing waters. However, he concluded that numeric standards should be different for fresh and saline waters because of different dieoff rates for the bacteria and viruses for differing salinity conditions.

Other studies examined additional illness symptoms associated with swimming in contaminated water, besides gastrointestinal illness, and identified other potentially useful bacterial indicators. Seyfried, et al. (1985), for example, examined swimming beaches in Toronto for respiratory illness, skin rashes, plus eye and ear problems, in addition to gastrointestinal illness. They found that total staphylococci correlated best with swimming associated total illness, plus ear, eye and skin illness. However, fecal streptococci and fecal coliforms also correlated (but not as well) with swimming associated total illness. Ferley, et al. (1989) examined illnesses among swimmers during the summer of 1986 in the French Ardèche river basin, during a time when untreated domestic sewage was entering the river. They examined total coliforms, fecal coliforms, fecal streptococci and Pseudomonas aeruginosa and Aeromonas Spp, but only two samples per week were available for each swimming area. The total morbidity rate ratio for swimmers compared to nonswimmers was 2.1 (with a 95% confidence interval of 1.8 to 2.4), with gastrointestinal illness the major illness observed. They found that fecal streptococci (FS) was the best indicator of gastrointestinal illness. A critical FS value of 20 MPN/100 mL indicated significant differences between the swimmers and nonswimmers. Skin ailments were also more common for swimmers than for nonswimmers and were well correlated with the concentrations of fecal coliforms, Aeromonas Spp and Pseudomonas aeruginosa. They noted that a large fraction (about 60%) of the fecal coliforms corresponded to E. coli, and that their definition of fecal streptococci essentially was what North American researchers termed enterococci.

Koenraad, et al. (1997) investigated the contamination of surface waters by Campylobacter and its associated human health risks. They reported that campylobacteriosis is one the most frequently occurring acute gastroenteritis diseases in humans. Typical investigations have focused on the consumption of poultry, raw milk, and untreated water as the major sources of this bacterial illness. Koenraad, et al. (1997) found that human exposures to Campylobacter contaminated surface waters is likely a more important risk factor than previously considered. In fact, they felt that Campylobacter infections may be more common than Salmonella infections. The incidence of campylobacteriosis due to exposure to contaminated recreational waters has been estimated to be between 1.2 to 170 per 100,000 individuals. The natural habitat of Campylobacter is the intestinal tract of warm-blooded animals (including poultry, pigs, cattle, gulls, geese, pigeons, magpies, rodents, shellfish, and even flies). It does not seem to multiply outside of its host, but it can survive fairly well in aquatic environments. It can remain culturable and infective for more than 2 months under ideal environmental conditions. Besides runoff, treated wastewater effluent is also a major likely source of Campylobacter in surface waters. Sanitary wastewater may contain up to 50,000 MPN of Campylobacter per 100 mL, with 90 to 99% reductions occurring during typical wastewater treatment.

Many of the available epidemiological studies have been confined to healthy adult swimmers, in relatively uncontaminated waters. However, it is assumed that those most at risk would be children, the elderly, and those chronically ill, especially in waters known to be degraded. Obviously, children are the most likely of this most-at-risk group to play in, or by, water. Alexander, et al. (1992) therefore specifically examined the risk of illness associated with swimming in contaminated sea water for children, aged 6 to 11 years old. This study was based on parental interviews for 703 child participants during the summer of 1990 at Blackpool beach, UK. Overall, 80% of the samples at the Blackpool Tower site and 93% of the samples at the South Pier site failed to meet the European Community Standards for recreational waters. All of the 11 designated beaches in Lancashire (including Blackpool beach), in the northwest region of England, continually fail the European directive imperative standards for recreational waters. During this study, statistically significant increases in disease were found for children who had water contact, compared to those who did not. Table XXXiv shows the prevalence and rate ratios for these symptoms. Diarrhea and loss of appetite had strong associations with the water contact group, while vomiting and itchy skin had moderate associations. No other variables examined (household income, sex of the child, sex of the respondent, general health, chronic or recurring illness in the child, age of the child, foods eaten, including ice cream, other dairy products, chicken, hamburgers, shellfish, or ice cubes, acute symptoms in other household members, presence of children under 5 in the household, and other swimming activities) could account for the significant increases in the reported symptoms for the children who had water contact.

Other risk factors, in addition to exposure to sewage contaminated swimming waters, was investigated by Fleisher, et al. (1993). People visiting beaches for recreation are frequently exposed to additional risks for gastroenteritis disease, especially related to foods that are eaten. Picnic lunches and food purchased at swimming beaches may contain improperly prepared or inadequately stored foods, including food that may be especially risky including sandwiches having mayonnaise, chicken, eggs, hamburgers, and hot dogs. They found that non-water related risk factors confounded the relationships between gastroenteritis and fecal streptococci densities. They also found that fecal coliform and fecal streptococci densities changed rapidly in time and location at swimming beaches, requiring many more water sample evaluations than are typically obtained during most epidemiological studies.

Swimming beach studies were conducted in Hong Kong during the summers of 1986 and 1987 (Cheung, et al. 1990). This was a significant study in that it was one of the first major epidemiological investigations that has been conducted in subtropical waters. The Hong Kong swimming beach criteria, adopted in 1981, set the following objective: "The level of E. coli should not exceed 1,000 per 100 mL, calculated as the running median of the most recent five consecutive samples." Beaches that did not meet this objective for 60% of the time in any year were closed to swimming.

The results of this study can be compared to the more common temperate area studies as an indication of the usability of recreation water quality criteria for a broader range of conditions. More than 18,700 responses were obtained from beachgoers on nine beaches. Water samples were collected every two hours at the nine beaches under study. The samples were analyzed for E. coli, Klebsiella spp., fecal streptococci, fecal coliforms, staphylococci, Pseudomonas aeruginosa, Candida albicans, and total fungi. E. coli only represented 57% of the fecal coliforms (much lower than reported elsewhere). Beachgoers were recruited on selected weekends and given initial interviews. Follow-up telephone interviews were obtained 7 to 10 days afterwards. The beachgoers spent an average of 3.5 hours at the beach, and swimmers spent an average of 1.3 hours in the water (much longer than reported in colder climates). The beaches studied were affected to varying degrees by nearby submarine sewage outfalls, agricultural runoff (pig farming) or by storm drains discharging across the beaches.

The overall symptom rates for gastrointestinal, ear, eye, skin, respiratory, fever, and total illness were significantly higher for swimmers than for non-swimmers. Many of the rates were also higher at "barely acceptable" beaches than at "relatively unpolluted" beaches. The increased risk of swimmers developing highly credible gastrointestinal illness (HCGI) was 5 times greater than for non-swimmers. The increased risk for swimmers in developing gastrointestinal (GI), eye, skin, and total illness was 2 to 4 times greater than for non-swimmers. The incubation period for the gastrointestinal symptoms in Hong Kong were similar to those reported for the U.S., indicating a possible similar causative agent (Norwalk virus and rotavirus virus originating from human sewage being suspected). Children under 10 years of age were also found to have significantly higher symptom rates for GI, HCGI, skin, respiratory, fever, and total illness than older swimmers. Escherichia coli was found to be the best indicator of swimmer illness (especially gastroenteritis and skin symptoms). Staphylococci measurements were recommended as a supplement to E. coli, especially for ear, respiratory and total illness. They contrasted this finding with typically better correlations between enterococci and health risks at U.S. beaches. They concluded that it may not be appropriate to adopt another country’s water contact recreation water quality criteria, especially if they are vastly separated geographically. Differences may be due to differences in the immune state of the populations and the indicator-illness relationships. Geometric mean densities of 180 E. coli per 100 mL and 1,000 staphylococci per 100 mL were found to be the thresholds for differentiating "barely acceptable" and "relatively unpolluted" beaches. These observations were used to develop new swimming beach standards for Hong Kong, as shown in Table XX10. This new classification scheme was in place in 1988.

Table XX10. Classification of Hong Kong Beaches Based on Swimming Associated Health Risk Levels

Cheung, et al. 1990.

This study examined problems associated with sewage contaminated swimming beaches (from CSO discharges and ocean outfalls of treated sewage) (Corbett, et al. 1993). They interviewed almost 3,000 beach goers at 12 beaches during 3 months in late 1989 and early 1990. Follow-up telephone interviews were conducted about a week later concerning incidence of illness. During the 41 days of sampling, 461 samples were analyzed for fecal coliforms and fecal streptococci. Of these samples, 67% failed to meet New South Wales Department of Health water quality criteria.

Swimmers were almost twice as likely as nonswimmers to report symptoms, but the prevalence of respiratory symptoms in people aged 15 to 25 was high, irrespective of swimming status or pollution level. The incidence of respiratory, fever, eye, ear, and other problems increased with increasing bacterial counts. Fecal streptococci counts were worse predictors of the swimming risk than the fecal coliform counts. Gastrointestinal symptoms were not related to either the fecal coliforms or fecal streptococci counts monitored. Those who swam for longer than 30 minutes were more than 4 times as likely to develop gastrointestinal symptoms compared to nonswimmers or those who swam for shorter periods. Luckily, children playing near and in urban streams are not likely to have such prolonged submerged exposures, and gastrointestinal problems may not be as serious as other water contact problems. The risk of respiratory, ear, and eye symptoms accounted wholly for the increases in illness observed. They reported that enteroviruses can cause respiratory symptoms and can persist in marine sediments and waters for many months.

Table XX7 shows the percentages of swimmers who reported various illness symptoms after swimming in waters having varying bacterial contamination levels. Increasing levels of contamination increased the health risks for all symptoms, except for gastrointestinal symptoms. Table XX8 shows the odds ratios (and associated 95% confidence intervals) for illness at different levels of fecal coliform contamination. Above 1,000 cfu/100 mL fecal coliforms, the associations for these illnesses are all strong, while they are at least moderate for all levels shown, compared to the nonswimmers. However, most of the confidence intervals were quite large, indicating large variability in the observations, as expected.

Another recent swimmer/sewage exposure study was conducted in the UK, reported by Kay, et al. (1994) and by Fleisher, et al. (1996). This study was unique in design and was able to develop dose-response relationships and critical exposure levels for a few illnesses associated with swimmer exposures to sewage contaminated waters. Adult volunteers (1528 study participants) were studied over four seasons from 1989 through 1992. After arriving at the beach, healthy volunteers were randomized into bather and nonbather groups with the duration and place of individual exposure being rigorously controlled. All of the study locations met European Community mandatory bacteriological marine bathing water quality criteria and were therefore not excessively contaminated.

The researchers found a clear dose-response relationship between increasing levels of fecal streptococci and increased risk of acquiring acute febrile respiratory illness. Only bathers exposed to the highest quartile of exposure (51 to 158 FS /100 mL) showed a statistically significant increase in risk compared to the non bathers. The odds ratio (OR) was 2.65 (moderate association), with a 95% confidence interval of 1.19 – 5.48 for acute fibrile respiratory illness and fecal streptococci. There was a clear dose-response relationship among the bathers. In addition, exposure to increased levels of fecal coliform organisms was found to be predictive of ear ailments among bathers. Figures XX1 and XX2 show the derived dose-response relationships for swimmers acquiring disease related to bacteria density in the swimming water.

Figure XX1. Bathers’ probability of acquiring acute febrile respiratory illness through exposure to increasing levels of fecal streptococci (Fleisher, et al. 1996).

Figure XX2. Bathers’ probability of acquiring ear infections through exposure to increasing levels of fecal coliforms (Fleisher, et al. 1996).

Thresholds of exposure to indicator organisms, below which bathers were at no excess risk of illness relative to nonbathers, were estimated to be 60 fecal streptococci organisms/100 mL for febrile respiratory illness and 100 fecal coliform organisms/100 mL for ear ailments. These threshold levels are quite low and are commonly exceeded in most urban streams. No dose-response relationships or threshold levels were found for any of the indicator organisms (total coliforms, fecal coliforms, fecal streptococci, total staphlococci and Pseudomonas aeruginosa) and eye or skin ailments. They concluded that the use of a single illness or indicator organism for establishing swimming criteria in marine waters is incorrect.

A recreational water quality criterion can be defined as a "quantifiable relationship between the density of an indicator in the water and the potential human health risks involved in the water's recreational use." From such a definition, a criterion can be adopted which establishes upper limits for densities of indicator bacteria in waters that are associated with acceptable health risks for swimmers.

The Environmental Protection Agency, in 1972, initiated a series of studies at marine and fresh water bathing beaches which were designed to determine if swimming in sewage-contaminated marine and fresh water carries a health risk for bathers; and, if so, to what type of illness. Additionally, the EPA wanted to determine which bacterial indicator is best correlated to swimming-associated health effects and if the relationship is strong enough to provide a criterion (EPA 1986: Ambient Water Quality Criteria for Bacteria - 1986, EPA 440/5-84-002, U.S. Environmental Protection Agency, Office of Water Regulations and Standards, Washington, DC., NTIS access #: PB 86-158-045).

Many of the above described U.S. studies were conducted as part of these EPA sponsored research activities. The quantitative relationships between the rates of swimming-associated health effects and bacterial indicator densities were determined using standard statistical procedures. The data for each summer season were analyzed by comparing the bacteria indicator density for a summer bathing season at each beach with the corresponding swimming-associated gastrointestinal illness rate for the same summer. The swimming-associated illness rate was determined by subtracting the gastrointestinal illness rate in nonswimmers from that for swimmers.

The EPA’s evaluation of the bacteriological data indicated that using the fecal coliform indicator group at the maximum geometric mean of 200 organisms per 100 mL, as recommended in Quality Criteria for Water would cause an estimated 8 illness per 1,000 swimmers at freshwater beaches.

Additional criteria, using E. coli and enterococci bacteria analyses, were developed using these currently accepted illness rates. These bacteria are assumed to be more specifically related to poorly treated human sewage than the fecal coliform bacteria indicator. The freshwater equations developed by Dufour (1984b) were used to calculate new indicator densities corresponding to the accepted gastrointestinal illness rates.

It should be noted that these indicators only relate to gastrointestinal illness, and not other problems associated with waters contaminated with other bacterial or viral pathogens. Common swimming beach problems associated with contamination by stormwater include skin and ear infections caused by Psuedomonas aeruginosa and Shigella.

National bacteria criteria have been established for contact with bacteria and are shown in Table 5.15. State standards usually also exist for fecal coliform bacteria. Typical public water supply standards (Alabama’s are shown) are as follows:

(i) Bacteria of the fecal coliform group shall not exceed a geometric mean of 2,000/100 mL; nor exceed a maximum of 4,000/100 mL in any sample. The geometric mean shall be calculated from no less than five samples collected at a given station over a 30-day period at intervals not less than 24 hours. The membrane filter counting procedure will be preferred, but the multiple tube technique (five-tube) is acceptable.

(ii) For incidental water contact and recreation during June through September, the bacterial quality of water is acceptable when a sanitary survey by the controlling health authorities reveals no source of dangerous pollution and when the geometric mean fecal coliform organism density does not exceed 100/100 mL in coastal waters and 200/100 mL in other waters. When the geometric mean fecal coliform organism density exceeds these levels, the bacterial water quality shall be considered acceptable only if a second detailed sanitary survey and evaluation discloses no significant public health risk in the use of such waters. Waters in the immediate vicinity of discharges of sewage or other wastes likely to contain bacteria harmful to humans, regardless of the degree of treatment afforded these wastes, are not acceptable for swimming or other whole body water-contact sports.

Standards for fish and wildlife waters are similar to the above standard for a public water supply, except part (i) has different limits: "Bacteria of the fecal coliform group shall not exceed a geometric mean of 1,000/100 mL on a monthly average value; nor exceed a maximum of 2,000/100 mL in any sample." Part (ii) is the same for both water beneficial uses.

The EPA full body contact recreation water quality criteria are as follows:

Marine waters: "Based on a statistically sufficient number of samples (generally not less than 5 samples equally spaced over a 30-day period), the geometric mean of the enterococci densities should not exceed 35 per 100 mL." (EPA 1986)

Fresh waters: "Based on a statistically sufficient number of samples (generally not less than 5 samples equally spaced over a 30-day period), the geometric mean of the bacterial densities should not exceed one or the other of the following (Note that only one indicator should be used. The regulatory agency should select the appropriate indicator for its conditions.):

E. coli, at a concentration of 126 per 100 mL, or

EPA 1986

Most of the comprehensive urban runoff studies that examined bacteria, and especially pathogens, were conducted in the late 1960s to the early 1980s. These early references are summarized in the following paragraphs to supplement the newer, more general references, listed elsewhere in this chapter.

As noted previously, the fecal coliform test is not specific for any one coliform type, or groups of types, but instead has an excellent positive correlation for coliform bacteria derived from the intestinal tract of warm blooded animals (Geldreich, et al. 1968). The fecal coliform test measures Escherichia coli as well as all other coliforms that can ferment lactose at 44.5^oC and are found in warm blooded fecal discharges. Geldreich (1976) found that the fecal coliform test represents over 96 percent of the coliforms derived from human feces and from 93 to 98 percent of those discharged in feces from other warm blooded animals, including livestock, poultry, cats, dogs, and rodents. The variations in the specific fecal coliform bacteria biotypes are related to both fecal moisture content and diet. Moisture and diet may also affect the variety of bacteria biotypes found in the fecal coliform populations from different animal groups. In many urban runoff studies, all of the fecal coliforms were E. coli (Quresh and Dutka 1979). Fecal strep. bacteria are all of the intestinal Streptococci bacteria from warm blooded animal feces (Geldreich and Kenner 1969). The types and concentrations of different bacteria biotypes varies for different animal sources. Qureshi and Dutka (1979) found that pathogenic bacteria biotypes are present in southern Ontario urban runoff and are probably from several different sources.

Van Donzel, et al. (1967) reviewed water-borne disease outbreak information for 1946 to 1960. Almost 26,000 cases were listed for almost 230 known outbreaks in the United States and Puerto Rico. At least 29 of these outbreaks, involving more than 9,000 cases, were associated with stormwater runoff caused by either runoff washing human and animal feces or sewage into wells, springs, streams, small reservoirs, and open water mains, or by widespread flooding of individual and public water systems. Several authors, however, did not think that urban runoff may present a significant health problem. Olivieri, et al. (1977a) did not believe that urban runoff constitutes a major health problem because of the large numbers of viable bacteria cells that must be consumed to establish an infection. For urban runoff, it may be impossible to consume enough bacteria cells to establish the infective dose.

Salmonella has been reported in some but not all urban stormwaters. Oureshi and Dutka (1979) frequently detected Salmonella in southern Ontario stormwaters. They did not find any predictable patterns of Salmonella isolations as they were found throughout the various sampling periods. Olivieri, et al. (1977a) found Salmonella frequently in Baltimore runoff, but at relatively low concentrations and required sample concentration. Typical concentrations were from five to 300 Salmonella organisms/ten liters. The concentrations of Salmonella were about ten times higher in the stormwater samples than in the urban stream receiving the runoff. They also did not find any marked seasonal variations in Salmonella concentrations. Field, et al. (1976) also stated that Salmonella were frequently found in most Baltimore urban runoff samples. Almost all of the stormwater samples that had fecal coliform concentrations greater that 2000 organisms/100 mL had detectable Salmonella concentrations. About 27 percent of the samples having fecal coliform concentrations less than 200 organisms/100 mL had detectable Salmonella.

However, quite a few urban runoff studies did not detect Salmonella. Schillinger and Stuart (1978) found that Salmonella isolations were not common in a Montana subdivision study and that the isolations did not correlate well with fecal coliform concentrations. Environment Canada (1980) stated that Salmonella were virtually absent from Ottawa storm drainage samples in 1979. They concluded that Salmonella are seldom present in significant numbers in Ottawa urban runoff. The types of Salmonella found in southern Ontario were S. thompson and S. typhimurium var copenhagen (Qureshi and Dutka, 1979).

Olivieri, et al. (1977b) stated that the primary human enteric disease producing Salmonella biotypes associated with the ingestion of water include S. typhi (typhoid fever), S. paratyphi (paratyphoid fever), and Salmonella species (salmonellosis). These biotypes are all rare except for Salmonella. The dose of Salmonella required to produce an infection is quite large (approximately 105 organisms). The salmonellosis health hazard associated with water contact in urban streams is believed to be small because of this relatively large infective dose. If two liters of stormwater having typical Salmonella concentrations (ten Salmonella organisms per/ten liters) is ingested, less than 0.001 of the required infective dose would be ingested. If a worse case Salmonella stormwater concentration of 10,000 organisms/ten liters occurred, the ingestion of 20 liters of stormwater would be necessary for an infective dose. They stated that the low concentrations of Salmonella, coupled with the unlikely event of consuming enough stormwater, make the Salmonella health hazard associated with urban runoff small.

Geldreich (1965) recommended a fecal coliform standard of 200 organisms/100 mL because the frequency of Salmonella detection increased sharply at fecal coliform concentrations greater than this value. Setmire and Bradford (1980) stated that the National Academy of Sciences recommends a fecal coliform standard of 70/100 mL in waters with shellfish harvesting to restrict Salmonella concentrations in edible tissues. Field, et al. (1976) concluded that the use of indicator bacteria to protect Salmonella ingestion is less meaningful in stormwater runoff than in other waters.

Marron and Senn (1974) pointed out the possibility of dogs transmitting salmonellosis. They did not feel that this constitutes a serious public health threat but people should be aware of the possibility of infection and direct contact with dog feces should be minimized.

Staphylococcus aureus is an important human pathogen as it can cause boils, carbuncles, abscesses, and impetigo on skin on contact. Olivieri, et al. (1977b) stated that the typical concentrations of Staphylococci are not very high in urban streams. They also stated that there was little information available relating the degree of risk of staph. infections with water concentrations. They concluded that Staph. aureus appears to be the most potentially hazardous pathogen associated with urban runoff, but there is no evidence available that skin, eye, or ear infections can be caused by the presence of this organism in recreational waters. They concluded that there is little reason for extensive public health concern over recreational waters receiving urban storm runoff containing staph. organisms.

Olivieri, et al. (1977b) stated that there is circumstantial evidence that Shigella is present in urban runoff and receiving waters and that it could present a significant health hazard. Shigella species causing bacillary dysentery are one of the primary human enteric disease producing bacteria agents present in water. The infective dose of Shigella necessary to cause dysentery is quite low (10 to 100 organisms). Because of this low required infective dose and the assumed presence of Shigella in urban waters, it may be a significant health hazard associated with urban runoff.

Streptococcus faecalis and atypical S. faecalis are of limited sanitary significance (Geldreich 1976). Streptococcus determinations in urban runoff are most useful for identifying the presence of S. bovis and S. equinus that are specific indicators of non-human, warm blooded animal pollution. However, it is difficult to interpret fecal strep. data when their concentrations are lower than 100 organisms/100 mL because of the ubiquitious occurrence of S. faecalis var. liquifaciens. This biotype is generally the predominant strep. biotype occurring at low fecal strep. concentrations.

Pseudomonas aeruginosa is reported to be the most abundant pathogenic bacteria organism in urban runoff and streams (Olivieri, et al. 1977b). This pathogen is associated with eye and ear infections and is resistant to antibiotics. P. aeruginosa concentrations in urban runoff are at significantly greater concentrations (about 100 items) than the values associated with most bathing beach studies. Cabelli, et al. (1976) stated that P. aeruginosa is indigenous in about 15 percent of the human population. Swimmer’s ear or other P. aeruginosa infections may, therefore, be caused by trauma to the ear canals associated with swimming and diving, and not exposure to P. aeruginosa in the bathing water.

Environment Canada (1980) stated that there is preliminary evidence of the direct relationship between very low levels of P. aeruginosa and an increase in incidents of ear infections in swimmers. They stated that a control level for this Pseudomonas biotype of between 23 and 30 organisms/100 mL was considered. Cabelli, et al. (1976) stated that P. aeruginosa densities greater than 10 organisms/100 mL were frequently associated with fecal coliform levels considerably less than 200 organisms/100 mL. P. aeruginosa densities were sometimes very low when the fecal coliform levels were greater than 200 organisms/100 mL. An average estimated P. aeruginosa density associated with a fecal coliform concentration of 200 organisms/100 mL is about 12/100 mL. They further stated that P. aeruginosa by itself cannot be used as a basis for water standards for the prevention of enteric diseases during recreational uses of surface waters. The determinations of this biotype should be used in conjunction with fecal coliform or other indicator organism concentrations for a specific location. They recommended that bathing beaches that are subject to urban runoff pollution be temporarily closed until the P. aeruginosa concentrations return to a baseline concentration.

Candida albicans is a yeast found in Ottawa area urban runoff and receiving waters (Environment Canada 1980). This yeast can cause oral, cutaneous, and vaginal mycosis. Other potential health problems associated with urban runoff might be from histoplas-mosis and cryptococcosis that are associated with accumulations of guano at various bird roosts in or near areas of human habitation (Locke 1974).

E. coli and Vibro cholerae are disease producing pathogens associated with the ingestion of water. The cholera pathogen is quite rare, but E. coli is more common in urban runoff. The required infective dose of both of these pathogens is about 108 organisms (Olivieri, et al. 1977b).

Dog feces are capable of transmitting many diseases, including leptospirosis, brucellosis, toxoplasmosis, tuberculosis and other diseases. However, these problems are quite rare and do not indicate a serious public health threat. Visceral larval migrans (VLM) is the most serious disease associated with dog feces. This mostly affects children under four years of age who ingest the bacteria through ingestion of feces or contaminated soil. Symptoms include blindness.

Viruses may also be important pathogens in urban runoff. Very small amounts of a virus are capable of producing infections or diseases, especially when compared to the large numbers of bacteria organisms required for infection (Berg 1965). The quantity of enterroviruses which must be ingested to produce infections is usually not known (Olivieri, et al. 1977b). Viruses are usually detected at low levels in urban receiving waters and storm runoff. They stated that even though the minimum infective doses may be small, the information available indicates that stormwater virus threats to human health is small. Because of the low levels of virus necessary for infection, dilution of viruses does not significantly reduce their hazard.

This study was the first large-scale epidemiological study in the U.S. to investigate possible adverse health effects associated with swimming in ocean waters affected by discharges from separate storm drains (SMBRP 1996). This was a follow-up study after previous investigations found that human fecal waste was present in the stormwater collection systems (Water Environment & Technology 1996b, Environmental Science & Technology 1996b, and Haile, et al. 1996).

During a four month period in the summer of 1995, about 15,000 ocean swimmers were interviewed on the beach and during telephone interviews one to two weeks later. They were queried concerning illnesses since their beach outing. The incidence of illness (such as fever, chills, ear discharge, vomiting, coughing with phlegm, and credible gastrointestinal illness) was significantly greater (from 44 to 127% increased incidence) for ocean goers who swam directly off the outfalls, compared to those who swam 400 yards away, as shown on Table XX4. As an example, the rate ratio (RR) for fever was 1.6, while it was 2.3 for ear discharges, and 2.2 for highly credible gastrointestinal illness comprised of vomiting and fever (HCGI). The approximated associations were weak for any of the symptoms, and moderate for the others listed. Disease incidence dropped significantly with distance from the storm drain. At 400 yards, and beyond, upcoast or downcoast, elevated disease risks were not found. The results did not change when adjusted for age, beach, gender, race, socioeconomic status, or worry about health risks associated with swimming at the beach.

These interviews were supplemented with indicator and pathogen bacteria and virus analyses in the waters. The greatest health problems were associated with times of highest concentrations (E. coli >320 cfu/100 mL, enterococcus > 106 cfu/100 mL, total coliforms >10,000 cfu/100 mL, and fecal coliforms > 400 cfu/100 mL). Bacteria populations greater than these are common in urban runoff and in urban receiving waters. Symptoms were found to be associated with swimming in areas where bacterial indicator levels were greater than these critical counts. Table XX5 shows the heath outcomes associated with swimming in areas having bacterial counts greater that these critical values. The association for enterococcus with bloody diarrhea was strong, and the association of total coliforms with skin rash was moderate, but nearly strong.

The ratio of total coliform to fecal coliform was found to be one of the better indicators for predicting health risks when swimming close to the storm drain. When the total coliforms were greater than 1,000 cfu/100 mL, the strongest effects were generally observed when the total to fecal coliform ratio was 2. The risks decreased as the ratio increased. In addition, illnesses were more common on days when enteric viruses were found in the water.

The percentage of survey days exceeding the critical bacterial counts were high, especially when closest to the storm drainage, as shown on Table XX6. High densities of E. coli, fecal coliforms and enterococcus were observed on more than 25% of the days, however, there was a significant amount of variability in observed counts in the water samples obtained directly in front of the drains. The variability and the frequency of high counts dropped considerable with distance from the storm drains. Upcoast bacteria densities were less than downcoast densities probably because of prevailing near-shore currents.

The SMBRP (1996) concluded that less than 2 miles of Santa Monica Bay’s 50 mile coastline had problematic health concerns due to the storm drains flowing into the Bay. They also concluded that the bacterial indicators currently being monitored do help predict risk. In addition, the total to fecal coliform ratio was found to be a useful additional indicator of illness. As an outcome of this study, the Los Angeles County Department of Health Services will post new warning signs advising against swimming near the outfalls ("Warning! Storm drain water may cause illness. No swimming"). These signs will be posted on both sides of all flowing storm drains in Los Angeles County. In addition, county lifeguards will attempt to warn and advise swimmers to stay away from areas directly in front of storm drain outlets, especially in ponded areas. The county is also accelerating their studies on sources of pathogens in stormwater.

In November of 1997, the State of California proposed new bacterial criteria for fresh and saltwater recreational areas (DHS 1997). These criteria are heavily based on the Santa Monica Bay study described above and recognize the danger that urban runoff presents. They recommend that recreational use of waters within stormwater drains (including manmade conveyances and also natural drains such as creeks and streams), in ponds or pools that form because of stormwater drainage, and in the immediate surf zone into which stormwater drains, should be prohibited at all times. The criteria documents state that:

Similar wording was also provided relating to swimming in freshwaters contaminated by urban runoff. Indicator organisms should include total and fecal coliform bacteria, at a minimum. Enterococci can also be added as an indicator. They felt that monitoring for specific pathogens (such as Giardia or Cryptosporidium) is costly and doesn’t appear to be reliable. They could be monitored if done in conjunction with the other required monitoring efforts, especially in response to specific needs. Levels indicating a need for additional attention (they suggested conducting sanitary surveys to identify and correct the sources of contamination) in both salt waters and freshwaters are:

Total coliforms: 1,000 per 100 mL (single sample), or

1,000 per 100 mL, in more than 20 percent of the samples at any sampling station, in any

Fecal coliforms: 200 per 100 mL, or

200 per 100 mL, based on the log mean of at least 5 equally spaced samples in a 30-day

In addition, when the local health officer considers enterococcus monitoring for supplemental information, the following levels are also recommended:

Enterococcus (salt water): 35 per 100 mL (single sample), or

35 per 100 mL, based on the log mean of at least 5 equally spaced

Enterococcus (freshwater): 33 per 100 mL (single sample), or

33 per 100 mL, based on the log mean of at least 5 equally spaced

Freshwater swimming areas could also be monitored for E. coli to provide additional supplemental information. In that case, the following level indicating a need for more attention is also provided:

E. coli: 126 per 100 mL (single sample), or

126 per 100 mL (log mean of samples over a 30-day period (EPA 1986)

Salt water beach closure is recommended when sampling indicates any of the following conditions, when confirmed within 24 to 48 hours:

Total coliforms: 10,000 per 100 mL (17 California Code of Regulations, Section 7958)

Total coliforms: 5,000 per 100 mL, if the coliform index (the ratio of fecal to total coliform counts, times

Fecal coliforms: 1,000 per 100 mL

When enterococcus monitoring is also used, the following closure level is recommended:

Enterococcus: 104 per 100 mL (EPA 1986)

Freshwater recreational areas should be closed whenever any of the following conditions are exceeded, when confirmed within 24 to 48 hours:

Total coliforms: 10,000 per 100 mL

Fecal coliforms: 400 per 100 mL (EPA 1986)

When enterococcus or E. coli monitoring is also used, the following closure level is recommended:

Enterococcus: 61 per 100 mL (EPA 1986)

E. coli: 235 per 100 mL (EPA 1986)

Reopening of a closed recreational area is appropriate when two successive samples taken at least 24 hours apart are below the closure levels. If a swimming area is closed due to contamination by urban stormwater runoff, the following wording for warning signs is suggested: "Warning! Closed to swimming. Beach/swimming area is contaminated by stormwater runoff/sewage and may cause illness." In areas that are chronically contaminated by stormwater, the following wording for permanent signs is suggested: "Warning! Storm drain water may cause illness. No swimming in storm drain water."

The National Research Council conducted an intensive review of the use of waters of impaired quality for groundwater recharge (Andelman, et al. 1994). Included in this book was a review of the use of stormwater to recharge groundwater for eventual use as a drinking water supply. Other potential source waters investigated for recharge included treated municipal wastewater and irrigation return flows. The following is a summary from that book, describing these potential human health risks associated with stormwater.

Various chemical and bacteriological health risks associated with stormwater were examined. The major risks were identified as originating from pathogenic organisms, disinfection byproducts for water that have undergone disinfection to reduce the threat from the pathogens, synthetic organic chemicals, and inorganic chemicals. Assessments are therefore needed to identify the potential risks associated with this reuse. These assessments contain four major components: hazard identification, dose-response assessment, exposure assessment, and risk characterization. The NRC committee reviewed available epidemiological studies that had investigated the use of degraded waters for recharge and as eventual drinking water supplies.

Table 4.2, summarized from the NRC report, lists the health effects of known chemicals found in urban stormwater. The health effects shown are not meant to be comprehensive, but are the problems that the drinking water standards are intended to protect against. The EPA carcinogen classifications are as follows:

A = sufficient evidence for humans

B1 = limited evidence for humans and sufficient evidence in experimental animals

B2 = inadequate/limited evidence for humans, sufficient evidence in experimental animals

C = limited evidence in experimental animals with no human data

D = inadequate or no data

E = sufficient evidence for noncarcinogenicity

The concentrations presented are summarized from the EPA’s Nationwide Urban Runoff Program (NURP) (EPA 1983) and show the percentage of samples where the toxicant was detected and the range of the detected values. Further information is given in Chapter 5 on stormwater characteristics. The maximum contaminant level (MCL) is the drinking water standard established by the EPA. Also shown (in parentheses) is the concentration associated with a cancer risk of 1 in a million, the generally recognized negligible risk level. The present background cancer occurrence rate in the U.S. is 25%. This 10^-6 risk level, associated with a lifetime exposure to a chemical, will increase the risk of getting cancer from 250,000 in 1 million to 250,001 in 1 million (Andelman, et al. 1994). The reference dose is the estimated daily dose that is likely to be without an appreciable risk of deleterious effects during a lifetime (expressed as mg of ingested chemical per day per kg of body weight).

Most of the listed toxicants exceed the MCL limits and the negligible risk levels (highlighted in bold). However, as will be shown in Chapter 5, most of the toxicants are associated with particulates and the MCL values are not directly applicable. In addition, drinking of undiluted, untreated stormwater is not likely.

Microorganisms of concern in drinking waters may include many different types of pathogens, including bacteria, viruses, and parasites. These are excreted from infected hosts and enter sanitary sewage. Stormwater and urban receiving waters can become contaminated with these pathogens, as noted earlier. Andelman, et al. (1994) reviewed waterborne disease outbreaks in the U.S. from 1971 through 1990. The most common identified causative agents were Giardia, chemical poisoning, and Shigella species. During this period, the causative agents in more than 50% of the outbreaks were not able to be identified. However, reviews of past outbreaks found that the Norwalk virus (causing acute nonbacterial gastroenteritis) was the likely cause of about 40% of the outbreaks from 1976 through 1980 that had no prior identified cause. The difficulty or inability to identify many of the viruses and parasites (such as Cryptosporidium) is the likely reason why they are not listed as a more common cause of illness from drinking contaminated water.

Dose-response information is usually determined by exposing volunteers to different doses of the microorganisms of interest. Normally, this data does not include special problems for special at-risk individuals. Table 4.8 (as reported in the NRC committee report) shows infective dose information for several pathogens. Table 4.12 shows the probability of infection of ingestion of 100 mL of water for various levels of contamination. As will be shown in Chapter 5, the levels of these microorganisms in stormwater can be much greater than the values shown on this table (enterviruses of 100 to 3000 pfu/100 L, for example was reported by Olivieri, et al. 1977). Of course, ingestion of untreated or undiluted stormwater is rare.

Craun, et al. (1997) conducted evaluations of waterborne disease outbreaks from public water supplies and found that coliform bacteria monitoring is likely adequate to protect against bacterial and viral illness, but not for protozoa associated illness. Coliform bacteria monitoring has been used for many years to assess the microbiological quality of drinking waters. Except for a few strains, coliforms are not considered pathogenic. They are not very specific to fecal contamination, as most species of coliforms are free-living in the environment. Tap water having no coliforms has generally been thought to be free of agents likely to cause waterborne disease. However, Craun, et al. (1997) found that disease outbreaks (especially associated with Giardia or Cryptosporidium) have occurred in water systems that have not violated the maximum contaminant levels for total coliforms. The 1989 Coliform Rule for drinking waters states that systems collecting fewer than 40 samples per month may have no more than one total coliform positive sample (per 100 mL of water) per month, systems collecting more samples must have fewer than 5% of their samples positive for total coliforms. When Craun, et al. (1997) reviewed information from reported waterborne disease outbreaks from 1983 to 1992, they found that coliforms were detected during most of the outbreaks that were caused by bacteria, viruses, and unidentified agents, but they were found only during few of the outbreaks caused by protozoa. As an example, the 1993 Milwaukee Cryptosporidium outbreak (the largest documented waterborne disease outbreak in the U.S., with 400,000 cases of illness reported) occurred even though the MCL for coliforms was not violated. It is known that total coliforms are more susceptible to disinfection during water treatment than some protozoa. They concluded that "microbiological monitoring alone (for total coliforms and other indicator organisms for pathogens) cannot safeguard the public against waterborne disease. Emphasis must also be given to source water protection (watershed control programs, better control of wastewater discharges, and wellhead protection programs) and adequate water treatment and operation. The 1989 coliform rule with its more stringent requirements (periodic sanitary surveys, procedures for E. coli testing, and extra samples to evaluate water quality after positive total coliform results) and other USEPA regulations (e.g. the Surface Water Treatment Rule, and the pending Enhanced Surface Water Treatment Rule) are all important for reducing the risks of waterborne disease."

Protozoa became an important public issue with the 1993 Cryptosporidium-caused disease outbreak in Milwaukee when about 400,000 people become ill from drinking contaminated water. Mac Kenzie, et al. (1994) prepared an overview of the outbreak, describing the investigation on the causes of the illness and the number of people affected. They point out that Cryptosporidium-caused disease in humans was first documented in 1976, but had received little attention and no routine monitoring. Cryptosporidium now is being monitored routinely at many areas and is the subject of much research concerning its sources and pathways. At the time of the Milwaukee outbreak, both of the city’s water treatment plants (using water from Lake Michigan) were operating within acceptable limits, based on required monitoring. However, at one of the plants (which delivered water to most of the infected people), the treated water experienced a large increase in turbidity (from about 0.3 NTU to about 1.5 NTU) at the time of the outbreak that was not being well monitored (the continuous monitoring equipment was not functioning, and values were only obtained every 8 hours). More than half of the residents receiving water from this plant became ill. The plant had recently changed its coagulant from polyaluminum chloride to alum and equipment to assist in determining the correct chemical dosages was not being used. The finished water had apparently relatively high levels of cryptosporidium because some individuals became ill after only drinking less than 1 L of water. Cryptosporidium oocysts have often been found in untreated surface waters, and it was thought that Cryptosporidium oocysts entered the water treatment supply before the increase in turbidity was apparent. Mac Kenzie, et al. (1994) point out that monitoring in the United Kingdom has uncovered sudden, irregular, community-wide increases in cryptosporidiosis that were likely caused by waterborne transmission. They also stated that the source of the Cryptosporidium oocysts was speculative, but could have included cattle feces contamination in the Milwaukee and Menomonee Rivers, slaughterhouse wastes, and human sewage. The rivers were also swelled by high spring rains and snowmelt runoff that may have aided the transport of upstream Cryptosporidium oocysts into the lake near the water intakes.

The Journal of the American Water Works Association has published numerous articles on protozoa contamination of drinking water supplies. Crockett and Haas (1997) describe a watershed investigation to identify sources of Giardia and Cryptosporidium in the Philadelphia watershed. They describe the difficulties associated with monitoring Cryptosporidium and Giardia in surface waters because of low analytical recoveries and the cost of analyses. Large variations in observed protozoa concentrations made it difficult to identify major sources during the preliminary stages of their investigations. They do expect that wastewater treatment plant discharges are a major local source, although animals (especially calves and lambs) are likely significant contributors. Combined sewer overflows had Giardia levels similar to raw sewage, but the CSOs were much less than the raw sewage for Cryptosporidium. LeChevallier, et al. (1997) investigated Giardia and Cryptosporidium in open reservoirs storing finished drinking water. This gave them an opportunity to observe small increases in oocyst concentrations associated from nonpoint sources of contamination from the highly controlled surrounding area. They observed significantly larger oocyst concentrations at the effluent (median values of 6.0 Giardia/100 L and 14 Cryptosporidium/100 L) in the reservoirs than in the influents (median values of 1.6 Giardia/100 L and 1.0 Cryptosporidium/100 L). No human wastes could influence any of the tested reservoirs and the increases were therefore likely caused by wastes from indigenous animals or birds, either directly contaminating the water, or through runoff from the adjacent wooded areas.

A Management Training Audioconference Seminar on Cryptosporidium and Water (MTA 1997) was broadcast in May of 1997 to familiarize state and local agencies about possible Cryptosporidium problems that may be evident after the EPA’s Information Collection Rule begins in July of 1997. This regulation will require all communities serving more than 100,000 people to monitor their source water for Cryptosporidium oocysts. If the source water has more than 10 Cryptosporidium oocysts per liter, then the finished water must also be monitored. It is likely that many source waters will be found to be affected by cryptosporidium. The reviewed one study that found the percentage of positive samples of Cryptosporidium in lakes, rivers, and springs was about 50 to 60% and about 5% in wells. In contrast, the percentage of samples testing positive for Giardia was about 10 to 20% in lakes and rivers, and very low in springs and wells.

Pfiesteria has a complex life cycle, including at lease 24 flagellated, amoeboid, and encysted stages. Only a few of these stages appears to be toxic, but their complex nature makes them difficult to identify by nonexperts (Maguire and Walker 1997). Pfiesteria spends much of its life span in a nontoxic predatory form, feeding on bacteria and algae, or as encysted dormant cells in muddy sediment. Large schools of oily fish (such as the Atlantic menhaden) trigger the encysted cells to emerge and excrete toxins. These toxins make the fish lethargic, so they remain in the area where the toxins attack the fish skin, causing open sores to develop. The Pfiesteria then feed on the sloughing fish tissue. Unfortunately, people working in the water during these toxin releases may also be affected (Zimmerman 1998).

Researchers suggest that excessive nutrients (causing eutrophication) increase the algae and other organic matter that the Pfiesteria and Atlantic menhaden use for food. The increased concentrations of Pfiesteria above natural background levels increase the likelihood of toxic problems. Maguire and Walker (1997) state that other factors apparently involved include stream hydraulics, water temperature, and salinity. They feel that Pfiesteria is only one example of the increasing threats affecting coastal ecosystems that are experiencing increased nutrient levels. Most of the resulting algal blooms only present nuisance conditions, but a small number can result in human health problems (mostly as shellfish poisonings). The increased nutrient discharges are mostly associated with agricultural operations, especially animal wastes from large poultry and swine operations. In the Pocomoke River watershed, the Maryland Department of Natural Resources estimates that about 80% of the phosphorus and 75% of the nitrogen load is from agricultural sources. Urban runoff may also be a causative factor of eutrophication in coastal communities, especially those having small enclosed coastal lagoons or embayments, or in rapidly growing urban areas. Zimmerman (1998) points out that the Chesapeake Bay area is one of the country’s most rapidly growing areas, with the population expected to increase by 12 percent by the year 2010.

Aesthetics has historically been difficult to quantify in urban receiving water studies, but it has been an important parameter for many uses, especially recreation. Heidtke and Tauriainen (1996) developed an aesthetic rating system for the Rouge River (Detroit, MI), using a combination of water clarity, water color, odor, and visible debris. Their preliminary work suggests that the index is an effective tool for tracking time and space trends in aesthetic characteristics of the receiving water and for public education. Mizutani (1996) described the changes that have occurred in Japanese streams with urbanization. He describes efforts to improve the urban stream conditions by: augmenting stream flows (by releases from stormwater detention basins or by using highly treated sanitary wastewater), and by improving the designs of the walls of the streams (which are commonly concrete box channels).

Prior to urbanization, groundwater recharge resulted from infiltration of precipitation through pervious surfaces, including grasslands and woods. This infiltrating water was relatively uncontaminated. With urbanization, the permeable soil surface area through which recharge by infiltration could occur was reduced. This resulted in much less groundwater recharge and greatly increased surface runoff. In addition, the waters available for recharge generally carried increased quantities of pollutants. With urbanization, new sources of groundwater recharge also occurred, including recharge from domestic septic tanks, percolation basins and industrial waste injection wells, and from agricultural and residential irrigation.

One of the major concerns of stormwater infiltration is the question of adversely impacting groundwater quality. Pitt, et al. (1996) reviewed many studies that investigated groundwater contamination from stormwater infiltration. They developed a methodology to evaluate the contamination potential of stormwater nutrients, pesticides, other organic compounds, pathogens, metals, salts and other dissolved minerals, suspended solids, and gases, based on the concentrations of the contaminant in stormwater, the treatability of the contaminant, and the mobility of the contaminant through the vadose. Stormwater salts, some pathogens, 1,3-dichlorobenzene, pyrene, fluoranthene, and zinc, were found to have high potentials for contaminating groundwater, under some conditions. They concluded that there is only minimal potential of contaminating groundwaters from residential area stormwaters (chlorides in northern areas remains a concern), especially if surface infiltration is used.

The following paragraphs (from Pitt, et al. 1994 and 1996) describe the stormwater pollutants that have the greatest potential of adversely affecting groundwater quality during inadvertent or intentional stormwater infiltration, along with suggestions on how to minimize these potential problems.

Urban pesticide contamination of groundwater can result from municipal and homeowner use of pesticides for pest control and their subsequent collection in stormwater runoff. Pesticides that have been found in urban groundwaters include: 2,4-D, 2,4,5-T, atrazine, chlordane, diazinon, ethion, malathion, methyl trithion, silvex, and simazine. Heavy repetitive use of mobile pesticides on irrigated and sandy soils likely contaminates groundwater. Fungicides and nematocides must be mobile in order to reach the target pest and hence, they generally have the highest contamination potential. Pesticide leaching depends on patterns of use, soil texture, total organic carbon content of the soil, pesticide persistence, and depth to the water table.

The greatest pesticide mobility occurs in areas with coarse-grained or sandy soils without a hardpan layer, having low clay and organic matter content and high permeability. Structural voids, which are generally found in the surface layer of finer-textured soils rich in clay, can transmit pesticides rapidly when the voids are filled with water and the adsorbing surfaces of the soil matrix are bypassed. In general, pesticides with low water solubilities, high octanol-water partitioning coefficients, and high carbon partitioning coefficients are less mobile. The slower moving pesticides have been recommended in areas of groundwater contamination concern. These include the fungicides iprodione and triadimefon, the insecticides isofenphos and chlorpyrifos and the herbicide glyphosate. The most mobile pesticides include: 2,4-D, acenaphthylene, alachlor, atrazine, cyanazine, dacthal, diazinon, dicamba, malathion, and metolachlor.

Pesticides decompose in soil and water, but the total decomposition time can range from days to years. Literature half-lives for pesticides generally apply to surface soils and do not account for the reduced microbial activity found deep in the vadose zone. Pesticides with a thirty-day half life can show considerable leaching. An order-of-magnitude difference in half-life results in a five- to ten-fold difference in percolation loss. Organophosphate pesticides are less persistent than organochlorine pesticides, but they also are not strongly adsorbed by the sediment and are likely to leach into the vadose zone, and the groundwater.

The most commonly occurring organic compounds that have been found in urban groundwaters include phthalate esters (especially bis(2-ethylhexyl)phthalate) and phenolic compounds. Other organics more rarely found, possibly due to losses during sample collection, have included the volatiles: benzene, chloroform, methylene chloride, trichloroethylene, tetrachloroethylene, toluene, and xylene. PAHs (especially benzo(a)anthracene, chrysene, anthracene and benzo(b)fluoroanthenene) have also been found in groundwaters near industrial sites.

Groundwater contamination from organics, like from other pollutants, occurs more readily in areas with sandy soils and where the water table is near the land surface. Removal of organics from the soil and recharge water can occur by one of three methods: volatilization, sorption, and degradation. Volatilization can significantly reduce the concentrations of the most volatile compounds in groundwater, but the rate of gas transfer from the soil to the air is usually limited by the presence of soil water. Hydrophobic sorption onto soil organic matter limits the mobility of less soluble base/neutral and acid extractable compounds through organic soils and the vadose zone. Sorption is not always a permanent removal mechanism, however. Organic re-solubilization can occur during wet periods following dry periods. Many organics can be at least partially degraded by microorganisms, but others cannot. Temperature, pH, moisture content, ion exchange capacity of soil, and air availability may limit the microbial degradation potential for even the most degradable organic.

Viruses have been detected in groundwater where stormwater recharge basins were located short distances above the aquifer. Enteric viruses are more resistant to environmental factors than enteric bacteria and they exhibit longer survival times in natural waters. They can occur in potable and marine waters in the absence of fecal coliforms. Enteroviruses are also more resistant to commonly used disinfectants than are indicator bacteria, and can occur in groundwater in the absence of indicator bacteria.

The factors that affect the survival of enteric bacteria and viruses in the soil include pH, antagonism from soil microflora, moisture content, temperature, sunlight, and organic matter. The two most important attributes of viruses that permit their long-term survival in the environment are their structure and very small size. These characteristics permit virus occlusion and protection within colloid-size particles. Viral adsorption is promoted by increasing cation concentration, decreasing pH and decreasing soluble organics. Since the movement of viruses through soil to groundwater occurs in the liquid phase and involves water movement and associated suspended virus particles, the distribution of viruses between the adsorbed and liquid phases determines the viral mass available for movement. Once the virus reaches the groundwater, it can travel laterally through the aquifer until it is either adsorbed or inactivated.

The major bacterial removal mechanisms in soil are straining at the soil surface and at intergrain contacts, sedimentation, sorption by soil particles, and inactivation. Because of their larger size than for viruses, most bacteria are therefore retained near the soil surface due to this straining effect. In general, enteric bacteria survive in soil between two and three months, although survival times up to five years have been documented.

Heavy metals and other inorganic compounds in stormwater of most environmental concern, from a groundwater pollution standpoint, are aluminum, arsenic, cadmium, chromium, copper, iron, lead, mercury, nickel, and zinc. However, the majority of these compounds, with the consistent exception of zinc, are mostly found associated with the particulate solids in stormwaters and are thus relatively easily removed through sedimentation practices. Filterable forms of the metals may also be removed by either sediment adsorption or are organically complexed with other particulates.

In general, studies of recharge basins receiving large metal loads found that most of the heavy metals are removed either in the basin sediment or in the vadose zone. Dissolved metal ions are removed from stormwater during infiltration mostly by adsorption onto the near-surface particles in the vadose zone, while the particulate metals are filtered out at the soil surface. Studies at recharge basins found that lead, zinc, cadmium, and copper accumulated at the soil surface with little downward movement over many years. However, nickel, chromium, and zinc concentrations have exceeded regulatory limits in the soils below a recharge area at a commercial site. Elevated groundwater heavy metal concentrations of aluminum, cadmium, copper, chromium, lead, and zinc have been found below stormwater infiltration devices where the groundwater pH has been acidic. Allowing percolation ponds to go dry between storms can be counterproductive to the removal of lead from the water during recharge. Apparently, the adsorption bonds between the sediment and the metals can be weakened during the drying period.

Similarities in water quality between runoff water and groundwater has shown that there is significant downward movement of copper and iron in sandy and loamy soils. However, arsenic, nickel, and lead did not significantly move downward through the soil to the groundwater. The exception to this was some downward movement of lead with the percolation water in sandy soils beneath stormwater recharge basins. Zinc, which is more soluble than iron, has been found in higher concentrations in groundwater than iron. The order of attenuation in the vadose zone from infiltrating stormwater is: zinc (most mobile) > lead > cadmium > manganese > copper > iron > chromium > nickel > aluminum (least mobile).

Salt applications for winter traffic safety is a common practice in many northern areas and the sodium and chloride, which are collected in the snowmelt, travel down through the vadose zone to the groundwater with little attenuation. Soil is not very effective at removing salts. Salts that are still in the percolation water after it travels through the vadose zone will contaminate the groundwater. Infiltration of stormwater has led to increases in sodium and chloride concentrations above background concentrations. Fertilizer and pesticide salts also accumulate in urban areas and can leach through the soil to the groundwater.

Studies of depth of pollutant penetration in soil have shown that sulfate and potassium concentrations decrease with depth, while sodium, calcium, bicarbonate, and chloride concentrations increase with depth. Once contamination with salts begin, the movement of salts into the groundwater can be rapid. The salt concentration may not decrease until the source of the salts is removed.

Table D-1 is a summary of the pollutants found in stormwater that may cause groundwater contamination problems for various reasons. This table does not consider the risk associated with using groundwater contaminated with these pollutants. Causes of concern include high mobility (low sorption potential) in the vadose zone, high abundance (high concentrations and high detection frequencies) in stormwater, and high soluble fractions (small fraction associated with particulates which would have little removal potential using conventional stormwater sedimentation controls) in the stormwater. The contamination potential is the lowest rating of the influencing factors. As an example, if no pretreatment was to be used before percolation through surface soils, the mobility and abundance criteria are most important. If a compound was mobile, but was in low abundance (such as for VOCs), then the groundwater contamination potential would be low. However, if the compound was mobile and was also in high abundance (such as for sodium chloride, in certain conditions), then the groundwater contamination would be high. If sedimentation pretreatment was to be used before infiltration, then much of the pollutants will likely be removed before infiltration. In this case, all three influencing factors (mobility, abundance in stormwater, and soluble fraction) would be considered important. As an example, chlordane would have a low contamination potential with sedimentation pretreatment, while it would have a moderate contamination potential if no pretreatment was used. In addition, if subsurface infiltration/injection was used instead of surface percolation, the compounds would most likely be more mobile, making the abundance criteria the most important, with some regard given to the filterable fraction information for operational considerations.

This table is only appropriate for initial estimates of contamination potential because of the simplifying assumptions made, such as the likely worst case mobility measures for sandy soils having low organic content. If the soil was clayey and had a high organic content, then most of the organic compounds would be less mobile than shown on this table. The abundance and filterable fraction information is generally

applicable for warm weather stormwater runoff at residential and commercial area outfalls. The concentrations and detection frequencies would likely be greater for critical source areas (especially vehicle service areas) and critical land uses (especially manufacturing industrial areas).

The stormwater pollutants of most concern (those that may have the greatest adverse impacts on groundwaters) include:

· nutrients: nitrate has a low to moderate groundwater contamination potential for both surface percolation and subsurface infiltration/injection practices because of its relatively low concentrations found in most stormwaters. However, if the stormwater nitrate concentration was high, then the groundwater contamination potential would also likely be high.

· pesticides: lindane and chlordane have moderate groundwater contamination potentials for surface percolation practices (with no pretreatment) and for subsurface injection (with minimal pretreatment). The groundwater contamination potentials for both of these compounds would likely be substantially reduced with adequate sedimentation pretreatment. Pesticides have been mostly found in urban runoff from residential areas, especially in dry-weather flows associated with landscaping irrigation runoff.

· other organics: 1,3-dichlorobenzene may have a high groundwater contamination potential for subsurface infiltration/injection (with minimal pretreatment). However, it would likely have a lower groundwater contamination potential for most surface percolation practices because of its relatively strong sorption to vadose zone soils. Both pyrene and fluoranthene would also likely have high groundwater contamination potentials for subsurface infiltration/injection practices, but lower contamination potentials for surface percolation practices because of their more limited mobility through the unsaturated zone (vadose zone). Others (including benzo(a)anthracene, bis (2-ethylhexyl) phthalate, pentachlorophenol, and phenanthrene) may also have moderate groundwater contamination potentials, if surface percolation with no pretreatment, or subsurface injection/infiltration is used. These compounds would have low groundwater contamination potentials if surface infiltration was used with sedimentation pretreatment. Volatile organic compounds (VOCs) may also have high groundwater contamination potentials if present in the stormwater (likely for some industrial and commercial facilities and vehicle service establishments). The other organics, especially the volatiles, are mostly found in industrial areas. The phthalates are found in all areas. The PAHs are also found in runoff from all areas, but they are in higher concentrations and occur more frequently in industrial areas.

· pathogens: enteroviruses likely have a high groundwater contamination potential for all percolation practices and subsurface infiltration/injection practices, depending on their presence in stormwater (likely if contaminated with sanitary sewage). Other pathogens, including Shigella, Pseudomonas aeruginosa, and various protozoa, would also have high groundwater contamination potentials if subsurface infiltration/injection practices are used without disinfection. If disinfection (especially by chlorine or ozone) is used, then disinfection byproducts (such as trihalomethanes or ozonated bromides) would have high groundwater contamination potentials. Pathogens are most likely associated with sanitary sewage contamination of storm drainage systems, but several bacterial pathogens are commonly found in surface runoff in residential areas.

· heavy metals: nickel and zinc would likely have high groundwater contamination potentials if subsurface infiltration/injection was used. Chromium and lead would have moderate groundwater contamination potentials for subsurface infiltration/injection practices. All metals would likely have low groundwater contamination potentials if surface infiltration was used with sedimentation pretreatment. Zinc is mostly found in roof runoff and other areas where galvanized metal comes into contact with rainwater.

· salts: chloride would likely have a high groundwater contamination potential in northern areas where road salts are used for traffic safety, irrespective of the pretreatment, infiltration or percolation practice used. Salts are at their greatest concentrations in snowmelt and early spring runoff in northern areas.

The Technical University of Denmark (Mikkelsen, et al. 1996a and 1996b) has been involved in a series of tests to examine the effects of stormwater infiltration on soil and groundwater quality. They found that heavy metals and PAHs present little groundwater contamination threat, if surface infiltration systems are used. However, they express concern about pesticides which are much more mobile. Squillace, et al. (1996) along with Zogorski, et al. (1996) presented information concerning stormwater and its potential as a source of groundwater MTBE contamination. Mull (1996) stated that traffic areas are the third most important source of groundwater contamination in Germany (after abandoned industrial sites and leaky sewers). The most important contaminants are chlorinated hydrocarbons, sulfate, organic compounds, and nitrates. Heavy metals are generally not an important groundwater contaminant because of their affinity for soils. Trauth and Xanthopoulus (1996) examined the long-term trends in groundwater quality at Karlsruhe, Germany. They found that the urban landuse is having a long-term influence on the groundwater quality. The concentration of many pollutants have increased by about 30 to 40% over 20 years. Hütter and Remmler (1996) describe a groundwater monitoring plan, including monitoring wells that were established during the construction of an infiltration trench for stormwater disposal in Dortmund, Germany. The worst case problem expected is with zinc, if the infiltration water has a pH value of 4.

The Coyote Creek study is an example of a comprehensive receiving water study that investigated the effects of stormwater on the biological conditions in an urban creek. This research project included many different biological, chemical, and physical parameters to quantify these effects. The project was conducted by Pitt and Bozeman (1982) from 1977 through 1982, with funding from the Storm and Combined Sewer Section of the U.S. Environmental Protection Agency. The objective of this three-year monitoring study was to evaluate the sources and impacts of urban runoff on water quality and biological conditions in Coyote Creek. The three major elements of this study included: (1) identifying and describing important sources of urban runoff pollutants, (2) describing the effects of those pollutants on water quality, sediment quality, aquatic organisms, and the creeks’ associated beneficial uses, and (3) assessing potential measures for controlling the problem pollutants in urban runoff. In many cases, very pronounced gradients of water and biological quality indicators were observed. Cause and effect relationships cannot be conclusively proven in a study such as this, the degradation of conditions in Coyote Creek may be due to several factors, including urban runoff, stream flows (both associated and not associated with urban runoff), and natural conditions (e.g., drought, stream gradient, groundwater infiltration, etc.). Information collected during this study implied that the effects of various urban runoff constituents, especially organics and heavy metals in the water and in the polluted sediment, may be responsible for much of the adverse biological conditions observed.

The project involved conducting field measurements, observations, sampling, and other studies of Coyote Creek from March 1977 through August 1980. The study focused on the urban reaches of Coyote Creek, extending from Lake Anderson to the confluence with Silver Creek. In this reach of Coyote Creek, there are few flow or pollutant contributions other than urban runoff. The sampling areas were selected such that each included a stretch of stream several hundred meters long, which met prescribed criteria for physical, biological, and chemical homogeneity.

The following parameters were typically examined at each sampling location:

Figure 30-1 is a map of the San Francisco Bay Area showing the location of the Coyote Creek watershed, while Figure 30-2 is a detailed map of the Coyote Creek watershed. The watershed itself is about 70 kilometers (45 miles) long, 15 kilometers (10 miles) wide, and contains about 80,000 hectares (200,000 acres). Nearly 15 percent of the watershed consisted of developed urban areas during the study period. Most of the urban development is located in the northwest portion of the watershed.

Figure 30-1. San Francisco Bay Area and the location of the Coyote Creek watershed.

Figure 30-2. Detailed map of the Coyote Creek watershed.

For much of its length, Coyote Creek flows northwesterly along the western edge of the watershed. Elevations in the watershed range from sea-level to nearly 920 meters (3,000 ft.). Figure 30-3 shows the elevations of the various major sampling locations. Near the San Jose urban area, the watershed can be characterized as a broad plain with rolling foothills to the east. A portion of the watershed (i.e., the narrow strip between Lake Anderson and the urban area) is used for light but productive agriculture. The upper reaches and the headwaters of Coyote Creek are in extremely rugged terrain, with slopes commonly exceeding 30 percent. These upper areas can be characterized as chaparral-covered hills and gullies in a fairly natural state; they receive little use by man. Much of this land is within the Henry Coe State Park; non-park land is used primarily for low-density cattle grazing.

Figure 30-3. Elevations of the major sampling locations.

Several major facilities have been built on Coyote Creek to provide flood control and groundwater recharge. The largest are the dams which contain man-made reservoirs: Lake Anderson and Coyote Lake. Discharges from these lakes are controlled by the Santa Clara Valley Water District. The major study area was located between the farthest downstream dam (Lake Anderson) and the first major confluence (where Coyote Creek meets Silver Creek, within the City of San Jose). Within this 39-kilometer (24-mile) study area, approximately 16 kilometers (10 miles) are urban and 23 kilometers (14 miles) are non-urban. Sampling stations were located in both the urban and non-urban reaches of the stream for comparison.

Average daily flows in the northern part of the creek during dry weather were typically less than 1.5 cubic meters per second (50 cfs). Major storm flows, however, approach 30 cubic meters per second (1,000 cfs). The flows in the northern part of the creek were controlled largely by the discharge from Lake Anderson and Coyote Lake.

The beginning of the project followed two years of severe drought. The first major rains occurred the previous November (1977), and seasonal rains that occurred during the study period were considered normal. Typical rainfall averaged 33 centimeters (13 inches) per year in the area below Lake Anderson and 50 to 71 centimeters (20 to 28 inches) per year in the watershed above Lake Anderson. During the drought, which preceded this study, rainfall was only about one-half of these amounts.

Coyote Creek is an important element of the Santa Clara Valley Water District’s groundwater recharge program. Several recharge basins have been established adjacent to the stream channel within the study area. Diversion channels withdraw water from Coyote Creek, route it into these large basins, and return it back into the creek, depending upon such factors as season, streamflow, and groundwater level.

There are an average of 0.6 to 3 storm drain outfalls per kilometer (1 to 5 per mile) along the urban reach of Coyote Creek that was studied. The outfalls ranged from 20 to 180 centimeters (8 to 70 inches) in diameter, but most are about 75 centimeters (30 inches) in diameter. The drainage area per outfall ranged from 2 to 320 hectares (5 to 800 acres), but most of the outfalls drained areas smaller than 40 hectares (100 acres).

Table 30-1 describes the drainage areas which cumulatively contribute runoff flows to selected monitoring stations. The urban area stations had about 3 to 5 percent (1,700 to 2,500 hectares or 4,000 to 6,000 acres) of their total drainage areas urbanized, whereas the non-urban area stations had less than 0.1 percent of their drainage areas urbanized. The three stations designated as Hellyer, Sylvandale, and Senter were transition stations (about 0.6 to 1.5 percent of their drainage areas were urbanized).

Table 31-1

Sampling took place during all months during the complete project period. As an example, the biological sampling stressed the spring and summer seasons of all project years, while the water column and sediment samples were conducted about monthly.

All water and sediment sampling was conducted manually using either plastic (HDPE) or glass wide-mouth bottles. Sediment core samples were obtained using a liquid carbon dioxide freezing core sampler. All water and sediment samples were comprised of at least six subsamples from the sampling location reach that were composited before analysis. The samples were then appropriately preserved and delivered to a commercial analytical laboratory for EPA-approved analyses.

Biological samples for lead and zinc bioaccumulation measurements (e.g. mosquito fish, filamentous algae, crayfish, cattail plant segments) were obtained at selected sampling stations during the routine fish sampling activities.

Fish were collected by seining and electroshocking representative pool and riffle habitats at 40 locations within the Coyote Creek system. Most of the collection efforts (conducted during the spring and summer of the different project years) were focused on the portion of Coyote Creek between Lake Anderson and the confluence of Silver Creek. However, to further define the species composition and distribution of fishes, additional samples were obtained from both the upper and lower reaches of Coyote Creek, as well as from several locations within major tributaries. Captured fishes were identified and counted. The total length and weight were recorded for each specimen. Where numerous individuals of a particular species were encountered, only length range and aggregate weight were recorded, along with any abnormalities.

Quantitative collections of benthic macroinvertebrates were made at nine locations in Coyote Creek. Benthic macroinvertebrate samples were collected from natural substrates (e.g., cobbles, gravel, sand) in both pool and riffle habitats by means of an Ekman dredge (sample area of 0.023m²) or a Surber sampler (sample area of 0.093m²). Additionally, artificial substrates were used at six sampling locations. These consisted of pairs of Hester-Dendy multi-plate samplers constructed of multiple, parallel plates of tempered hardboard (sample area of 0.120m²). The Hester-Dendy samplers were left in riffle sections of the stream for 8 weeks and then removed and examined in the laboratory.

Qualitative benthic collections were also made with the use of a D-frame sweep net at all biological monitoring stations. The benthic samples were washed through a sieve having a mesh size of 500 mm. Organisms retained on the screen were removed and preserved in 10 percent formalin, transferred to 70 percent ethanol, identified to the lowest practicable taxon, and enumerated.

Attached algae samples were obtained from both natural and artificial substrates throughout the various reaches of Coyote Creek. Qualitative samples of attached algae were collected by scraping uniform areas of natural substrates such as logs and rocks. Quantitative collections of attached algae were made with the use of artificial substrates consisting of diatometers equipped with glass slides. These were suspended in the water column at six locations within the study area for eight weeks, then removed and examined in the laboratory.

Rooted aquatic plants were sampled qualitatively whenever they were encountered in the study area. Plant specimens were collected, pressed or preserved, and identified.

Table 30-2

Water quality upstream of the urbanized area was fairly consistent from site to site, but the quality changed markedly as the creek passed through the urbanized area. The water quality within the urbanized reach was generally poorer than at the stations upstream of the urban area. Similar differences between wet and dry weather were also noted for the non-urban area. However, the wet weather concentrations were typically much higher in the urban area than in the non-urban area. Several other constituents were also found in higher concentrations in the urban area than in the non-urban area during wet weather. Lead concentrations were more than 7 times greater in the urban reach than in the non-urban reach during dry weather. Nitrite concentrations were almost 7 times greater in the urban area. Ammonia nitrogen values in the urban area were 2.8 times greater than in the nonurban area. Other significant increases in urban area concentrations included chloride, nitrate, orthophosphate, COD, specific conductance, sulfate, and zinc. Conversely, the dissolved oxygen measurements were about 20 percent less in the urban reach than in the non-urban reach of the creek.

Selected water and sediment samples from the urban area reaches of Coyote Creek were analyzed as part of a nationwide screening effort to assess priority pollutant concentrations in urban runoff and urban receiving waters. Three samples were collected in January 1979, during a major storm. These included a runoff sample and samples of sediment and water from Coyote Creek. The sampling was conducted in and near the Martha Street outfall, which is located in a heavily urbanized area. Only 18 of the approximately 120 priority pollutants analyzed, were detected (base-neutrals: fluoranthene, diethyl phthalate, di-n-butyl phthalate, bis (2-ethyl hexyl) phthalate, anthracene, phenanthrene, and pyrene; the phenols: 2,4,6-tricholorphenol, 2,4-dimethylphenol, pentachlorophenol, and phenol; and heavy metals: arsenic, cadmium, copper, lead, mercury, and zinc). These priority pollutants that were found in Coyote Creek are generally the same as those found in most other urban runoff and receiving water samples collected nationwide (EPA 1983, Pitt and Barron 1990).

Sediment samples were collected at the major sampling locations three times during the study. Table 30-3 summarizes all of the Coyote Creek sediment quality measurements obtained during the entire project. Orthophosphates, TOC, BOD₅, sulfates, sulfur, and lead were all found in higher concentrations in the sediments from the urban area stations, as compared with those from the upstream, non-urban area stations. The median sediment particle sizes were also found to be significantly smaller at the urban area stations, reflecting a higher silt content. Sulfur, lead, and arsenic were found in substantially greater concentrations (4 to 60 times greater) for the urban area sediments compared to the non-urban area sediments.

Table 30-3

When all of the sediment data from the three monitoring periods were combined, very little difference was found to exist between the urban and non-urban area values for COD, total phosphate, arsenic, and median particle size. However, seasonal variations were found to be important. When the data from just one sampling period were considered alone, greater and more significant variations in constituent concentrations between the two reaches were observed.

Lead concentrations in the urban area sediments were markedly greater than those from the non-urban area, by a factor of about 6 times (which is the widest margin for any constituent monitored). Large differences were also found between the urban and non-urban area data for both sulfate and phosphate. Average zinc concentrations in the sediments were found to increase by only about 1.5 times, but with a high degree of confidence.

The largest difference between urban and non-urban area sediment (mg/kg) to water (mg/L) concentration ratios (S/W) was for lead, where the S/W ratio was over 3,000 for the urban area and only about 400 for the non-urban area. The total Kjeldahl nitrogen S/W ratio was about 5,500 for the urban area but exceeded 22,000 for the non-urban area. For the other constituents studied, the differences between the urban and non-urban area S/W ratios was much less. Lead, zinc, arsenic, and total Kjeldahl nitrogen all had S/W ratios of between 2,000 and 5,000 in the urban area. COD and total phosphate had S/W ratios of 1,300 and 670 respectively, while orthophosphate and sulfate had S/W ratios of only about 20 and 6, respectively.

Because of these high observed sediment pollutant concentrations, it is likely that urban runoff affected sediment is an important factor in the general decline in biological quality as Coyote Creek passes through the San Jose urban area. Other natural factors (e.g., stream gradient, temperature, and velocity changes) also probably contribute to this decline. For example, relatively flat creek gradients in the urban reach lead to low velocities which, in turn, encourage sedimentation of polluted particulates and allow temperatures to rise. Decreased flows in the urban area (due to diversions and infiltration) are an additional cause for changes in flow regime, water quality, and biological conditions.

Biological samples were collected from six stations in Coyote Creek and were analyzed to determine the lead and zinc they had accumulated while living in the creek. This sampling program was restricted to a single collection of organisms, with representative samples obtained from throughout the urban and non-urban stretches of the creek. Fish (Gambusia affinis), filamentous algae (Cladophora sp.), crayfish (Procambarus clarkii), and cattail plant segments (Typha sp.) were collected for analysis. An effort was made to collect similar specimens of the same species from each sampling location. All samples were rinsed to remove adhering sediment and were then chemically digested and analyzed for total lead and zinc content.

Some evidence of bioaccumulation of lead and zinc was found in many of the samples of algae, crayfish, and cattails. The measured concentrations of these metals in organisms (mg/kg) exceeded concentrations in the sediments (mg/kg) by up to a maximum factor of about 6). Concentrations of lead and zinc in the organisms exceeded water column concentrations by factors of 100 to 500 times, depending upon the organism. Lead concentrations in urban area samples of algae, crayfish, and cattails were found to be two to three times as high as in non-urban area samples (Table 30-4), whereas zinc concentrations in urban area algae and cattail samples were about three times as high as the concentrations in the samples from the non-urban areas (Table 30-5). Lead and zinc concentrations in fish tissue were not significantly different between the urban and non-urban area samples.

Table 30-4

Table 30-5

Several early studies examined bioaccumulation in urban environments (Wilber and Hunter 1980, Neff, et al. 1978, Phillips and Russo 1978, Ray and While 1976, Rolfe, et al. 1977, Spehan, et al. 1978, EPA 1978, and EPA 1980). The lead concentrations in Coyote Creek waters are probably lower than the critical levels necessary to cause significant bioaccumulation in most aquatic organisms. The whole body concentrations of zinc for the fish and crayfish were greater than many of the whole-body concentration values reported in the literature. The zinc concentrations in the Coyote Creek plants, however, were smaller than concentrations reported elsewhere for polluted waters.

Table 30-5

Introduced fishes often cause radical changes in the nature of the fish fauna present in a given waterbody or drainage system. In many cases they become the dominant fishes because they are able to out-compete the native fishes for food or space, or they may possess greater tolerance to environmental stress. In general, introduced species are most abundant in aquatic habitats modified by man while native fishes tend to persist mostly in undisturbed areas (Moyle and Nichols 1973). Such was apparently the case within Coyote Creek. As seen on Table 30-6, samples from the non-urban portion of the study area were dominated by an assemblage of native fish species such as hitch, threespine stickleback, Sacramento sucker and prickly sculpin. Collectively, native species comprised 89 percent of the number and 79 percent of the biomass of the 2,379 fishes collected from the upper reaches of the study area. In contrast, native species accounted for only 7 percent of the number and 31 percent of the biomass of the 2,899 fishes collected from the urban reach of the study area.

Table 30-6

Hitch was the most numerous native fish species present. Hitch generally exhibit a preference for quiet water habitat and are characteristic of warm, low elevation lakes, sloughs, sluggish rivers and ponds (Calhoun 1966, Moyle and Nichols 1976). In streams of the San Joaquin River system in the Sierra Nevada foothills of central California, Moyle and Nichols (1973) found hitch to be most abundant in warm, sandy-bottomed streams with large pools, where introduced species such as green sunfish, largemouth bass, and mosquitofish were common. Likewise, during this Coyote Creek study, hitch were found to be associated with green sunfish, fathead minnows, and mosquitofish in the lower portions of Coyote Creek. However, mosquitofish dominated the collections from the urbanized section of the creek and accounted for over two-thirds of the total number of fish collected from that area. In foothill streams of the Sierra Nevada, Moyle and Nichols (1973) found mosquitofish to be most abundant in disturbed portions of the intermittent streams, especially in warm, turbid pools. The fish is particularly well adapted to withstand extreme environmental conditions, including those imposed by stagnant waters with low dissolved oxygen concentrations and elevated temperature. The second most abundant fish species in the urbanized reach of Coyote Creek, the fathead minnow, is equally well suited to tolerate extreme environmental conditions. The species can withstand low dissolved oxygen, high temperature, high organic pollution and high alkalinities. Often thriving in unstable environments such as intermittent streams, the fathead minnow can survive in a wide variety of habitats. However, the species seems to do best in pools of small, muddy streams and in ponds (Moyle and Nichols 1976).

The taxonomic composition and relative abundance of benthic macroinvertebrates were collected from both natural and artificial substrates in Coyote Creek. The abundance and diversity of benthic taxa were greatest in the non-urbanized sections of the stream. Figure 30-4 shows the trend of the overall decrease in the total number of benthic taxa encountered in the urbanized sections of the study area during 1978 and 1979. An overall increase in number and diversity of benthic organisms were encountered in 1979, compared to 1978 collections. This may be attributed to further recovery from the drought conditions that preceded this study. The benthos in the upper reaches of Coyote Creek consisted primarily of amphipods and a diverse assemblage of aquatic insects. Together those groups comprised two thirds of the benthos collected from the non-urban portion of the creek. Clean water forms were abundant and included amphipods (Hyaella azteca) and various genera of mayflies, caddisflies, black flies, crane flies, alderflies, and riffle beetles. In contrast, the benthos of the urban reaches of the creek consisted almost exclusively of pollution tolerant oligochaete worms (tubificids). Tubificids accounted for 97 percent of the benthos collected from the lower portion of Coyote Creek.

Figure 30-4 Trend of total number of benthic taxa observed during 1978 and 1979.

Crayfish were present throughout the study area and were collected in conjunction with the fish sampling effort. Two species of crayfish were encountered in Coyote Creek waters--Pacifastacus leniusculus and Procambarus clarkii. Neither species is native to California waters. Pacifastacus leniusculus was collected in the non-urbanized section of the study area. It is typically found in a wide variety of habitats including large rivers, swift or sluggish streams, lakes, and, occasionally, muddy sloughs (Rigel 1959). Procambarus clarkii was collected in both the urbanized and nonurbanized sections of the stream. Riegel (1959) states that the species prefers sloughs where the water is relatively warm and vegetation plentiful; however, it is also found in large streams. Because of its burrowing activities Procombarus clarkii often becomes a nuisance by damaging irrigation ditches and earthen dams.

Qualitative samples from natural substrates indicated that the filamentous alga, Cladophora sp. was found throughout the study area. However, its growth reached greatest proportions in the upper sections of the stream. Table 30-7 presents the taxonomic composition and relative abundance of diatoms collected from artificial substrates placed at selected sample locations. The periphyton of the non-urban reaches of the stream was dominated by the genera Cocconeis and Achnanthes. The genera Nitzschia and Navicula, generally accepted to be more pollution-tolerant forms, dominated the periphyton of the urbanized reaches of Coyote Creek.

Table 30-7

Rooted aquatics were not greatly abundant in the Coyote Creek study area. Submerged macrophytes were restricted entirely to the upper reaches of the study area and consisted of occasional stands of sago pondweed (Potamogeton pectinatus) and curly-leaf pondweed (Potamogeton crispus). Emergent forms consisted of water primrose (Jussiaea sp.), confined to several areas in the non-urban reach of the stream, and numerous small stands of cattails (Typha sp.) sparsely distributed throughout the length of the study area.

The biological investigations in Coyote Creek indicated distinct differences in the taxonomic composition and relative abundance of the aquatic biota present in Coyote Creek. The non-urban sections of the creek supported a comparatively diverse assemblage of aquatic organisms including an abundance of native fishes and numerous benthic macroinvertebrate taxa. In contrast, however, the urban portions of the creek comprised an aquatic community generally lacking in diversity and was dominated by pollution-tolerant organisms such as mosquitofish and tubificid worms.

Although certain differences in physical habitat occurred in the downstream reaches of the study area (e.g., a decrease in stream gradient, shorter riffles, wider, deeper pools, etc.), such differences were not thought to be responsible for the magnitude of change noted in the aquatic biota of the urban reach of Coyote Creek.

Urban runoff monitoring during this project showed that stormwater was the significant contributor to the high levels of many toxic materials in the receiving water and sediments of the stream. In addition, changes in the nature of the stream substrate occurred as a result of the deposition of silt and debris which largely originate from urban runoff. Such changes were likely the primary reason for the decline in species abundance and diversity observed in the urban reaches of Coyote Creek.

The effects of urban runoff on receiving water aquatic organisms or other beneficial uses is very site specific. Different land development practices create substantially different runoff flow characteristics. Different rain patterns cause different particulate washoff, transport and dilution conditions. Local attitudes also define specific beneficial uses and, therefore, current problems. There is also a wide variety of water types receiving urban runoff, and these waters all have watersheds that are urbanized to various degrees. Therefore, it is not surprising that urban runoff effects, though generally dramatic, are also quite variable and site specific.

Previous attempts to identify urban runoff problems using existing water quality data have not been conclusive because of differences in sampling procedures and the common practice of pooling data from various sites, or conditions (Pitt 1991). It is therefore necessary to carefully design comprehensive, long-term studies to investigate urban runoff problems on a site-specific basis. Sediment transport, deposition, and chemistry play key roles in urban receiving waters and need additional research. Receiving water aquatic biological conditions, especially compared to unaffected receiving waters, should be studied as a supplement to laboratory bioassays. In-stream taxonomic surveys are sensitive to natural variations of pollutant concentrations, flows, and other habitat affects. However, laboratory studies are necessary to help understand potential cause and effect relationships because of their ability to better control exposure variables.

These specific studies need to examine beneficial uses directly, and not rely on published water quality criteria and water column measurements alone. Published criteria are usually not applicable to urban runoff because of the slug nature of urban runoff and the unique chemical speciation of its components. Typical natural water pollutant characteristics (especially chemical mixtures and exposure pulses) are difficult to interpret, compared to simpler artificial systems having continuous discharges of more uniform characteristics.

The long-term aquatic life effects of urban runoff are probably more important than short-term effects associated with specific events, and are related to site specific conditions associated with dilution, size of the watershed, and size of the stream. The long-term effects are probably related to habitat degradation, deposition and accumulation of toxic sediments, or the inability of the aquatic organisms to adjust to repeated exposures to high concentrations of toxic materials or high flow rates.

Abernathy, A.R. Oxygen Consuming Organics in Nonpoint Source Runoff. Office of Research and Development. U.S Environmental Protection Agency. EPA-600/3-81-033. PB81-205981. Corvallis, Oregon. May 1981.

Adams, S.M., L.R. Shugart, and G.R. Southworth. "Application of bioindicators in assessing the health of fish populations experiencing contaminant stress." In: Biomarkers of Environmental Contamination. Lewis Publishers. Ann Arbor, Michigan. 1990.

Alden, R.W. and L.W. Hall. "Geographic targeting of ambient toxicity in the Chesapeake Bay watershed." Abstract Book: SETAC 17^th Annual Meeting. pg. 104. Washington, D.C., Nov. 17 - 21, 1996.

Alexander, L.M., A. Heaven, A. Tennant, and R. Morris. "Symptomatology of children in contact with sea water contaminated with sewage." Journal of Epidemiology and Community Health. Vol. 46, pp. 340-344. 1992.

Allen, H.E. (editor). Metal Contaminated Aquatic Sediments. Ann Arbor Press. Chelsea, MI. 350 pgs. 1996.

Andelman, J., H. Bauwer, R. Charbeneau, R. Christman, J. Crook, A. Fan, D. Fort, W. Gardner, W. Jury, D. Miller, R. Pitt, G. Robeck, H. Vaux, Jr., J. Vecchioli, and M. Yates. Ground Water Recharge Using Waters of Impaired Quality. Committee on Ground Water Recharge. Water Science and Technology Board. National Research Council. National Academy Press. Washington, D.C. 1994.

Armstrong, J.W., R.M. Thom and K.K. Chew. "Impact of a combined sewer outfall on the abundance, distribution and community structure of subtital benthos." Marine Env. Res. Vol. 4. pp. 3-23. 1981.

Arnbjerg-Nielsen, S.H. and P. Harremoës. "Modelling of tipping bucket rain gauges: Single rain events and rain series." 7^th International Conference on Urban Storm Drainage. Hannover, Germany. Sept. 9-13, 1996. Edited by F. Sieker and H-R. Verworn. IAHR/IAWQ. SuG-Verlagsgesellschaft. Hannover, Germany. 1996.

Arhelger, M., P. Jensen. Amd Y-C. Su. "Houston ship channel sediments and their relation to water quality." Volume 4, pg. 329 - 337. WEFTEC’96: Proceedings of the 69^th Annual Conference & Exposition. Dallas, Texas. 1996.

Ball, J.E., R. Jenks, and D. Auborg. "Dry weather build-up of constituents on road surfaces." 7^th International Conference on Urban Storm Drainage. Hannover, Germany. Sept. 9-13, 1996. Edited by F. Sieker and H-R. Verworn. IAHR/IAWQ. SuG-Verlagsgesellschaft. Hannover, Germany. pp. 785-790. 1996.

Bannerman, R.T., A.D. Legg, and S.R. Greb. Quality of Wisconsin Stormwater, 1989-94. U.S. Geological Survey and Wisconsin Department of Natural Resources. USGS Open-file report 96-458. 26 pgs. Madison, WI. 1996.

Barbour, M.T. "Measuring the health of aquatic ecosystems using biological assessment techniques: A national perspective." Presented at the Effects of Watershed Developments and Management on Aquatic Ecosystems conference. Snowbird, UT, August 4 – 9, 1996. Edited by L.A. Roesner. ASCE, New York, pp. 18 – 33. 1997.

Barr Engineering Co. Management Alternatives on the Diagnostic Feasibility Study for Medicine Lake, Hennepin, County, Minnesota. PB89-210009. National Technical Information Service, Springfield, VA. April 1987.

Bartkowska, I. and A. Królikowski. "Quality evaluations of storm water." 7^th International Conference on Urban Storm Drainage. Hannover, Germany. Sept. 9-13, 1996. Edited by F. Sieker and H-R. Verworn. IAHR/IAWQ. SuG-Verlagsgesellschaft. Hannover, Germany. pp. 43-48. 1996.

Bartel, R.L., A.E. Maristany. "Wetlands and stormwater management: A case study of Lake Munson. Part II: Impacts on sediment and water quality." In: Wetlands: Concerns and Successes. Proceedings published by the American Water Resources Association. pp. 231-246. Bethesda, MA. 1989.

Bascombe, A.D. Urban Pollution: Biological Monitoring of Benthic Invertebrates for the Assessment of Heavy Metal Pollution in Urban Rivers. Middlesex Polytechnic, Urban Pollution Research Center. Report No.: RR-11. Enfield, England. March 1988.

Bascombe, A.D., J.B. Ellis, D.M. Revitt, and R.B.E. Shutes. "Development of ecotoxicological criteria in urban catchments." Water Science and Technology. Vol. 22, No. 10/1, pp. 173-179. 1990.

Baudo, R., and H. Muntau. "Lesser known in-place pollutants and diffuse source problems." In: Sediments: Chemistry and Toxicity of In-Place Pollutants. Lewis Publishers. Ann Arbor, Michigan, 1990.

Baudo, R., J.P. Giesy, and H.Muntau. Sediments: Chemistry and Toxicity of In-Place Pollutants. Lewis Publishers, Ann Arbor, Michigan, 1990.

Belding, D.L. "Toxicity experiments with fish in reference to trade waste pollution." Trans. American Fisheries Society. Vol. 57. pp. 100-119. 1927.

Bellinger, E.G. "The response of algal populations to changes in lake water quality." In: Biological Indicators of Water Quality. John Wiley and Sons. New York. 1979.

Benke, A.C., G.E. Willeke, F.K. Parrish,and D.L. Stites. Effects of Urbanization on Stream Ecosystems. School of Biology, Environmental Resources Center, Report No. ERC 07-81. Georgia Institute of Technology. Atlanta, Georgia. 1981.

Benson, W.H., K.N. Baer, and C.F. Watson. "Metallothionein as a biomarker of environmental metal contamination." In: Biomarkers of Environmental Contamination. Lewis Publishers. Ann Arbor, Michigan. 1990.

Berg, G., editor. Transmission of Viruses by the Water Route. Interscience Publishers, NY. 1965.

Beyer, D.L., P.A. Kingsbury, and J.E. Butts. History and Current Status of Water Quality and Aquatic Ecology Studies in the Lower Chehalis River and Grays Harbor, Washington. Prepared for Washington Public Power Supply System. 1979.

Birtwell, I.K., C.D. Levings, J.S. MacDonal, and I.H. Rogers. "A review of fish issues in the Fraser River system." Journal Canadian Water Pollution Research. Vol. 23, No. 1, pp. 1-30. 1988.

Bjornn, T.C. Sediment in Streams and its Effects on Aquatic Life. PB-238594. U.S. Environmental Protection Agency. Washington, D.C. October 1974.

Blodgett, K.D., R.E. Sparks, A.A. Paparo, R.A. Cahill, and R.V. Anderson. "Distribution of toxicity in the sediments of the Illinois waterway." In: Urban Effects on Water Quality and Quantity Conference Proceedings. Illinois Environmental Protection Agency. Document No. 84/06. Springfield, Illinois. May 1984.

Bolstad, P.V. and W.T. Swank. "Cumulative impacts of landuse on water quality in a southern Appalachian watershed." Journal of the American Water Resources Association. Vol. 33, no. 3, pp. 519 – 533. June 1997.

Booth, D.B., D.R. Montgomery, and J. Bothel. "Large woody debris in urban streams of the Pacific Northwest" Presented at the Effects of Watershed Developments and Management on Aquatic Ecosystems conference. Snowbird, UT, August 4 – 9, 1996. Edited by L.A. Roesner. ASCE, New York, 1997.

Booth, D.B., C.R. Jackson. "Urbanization of aquatic systems: Degradation thresholds, stormwater detection, and the limits of mitigation." Journal of the American Water Resources Association. Vol. 33, no. 5, pp. 1077 – 1090. October 1997.

Borchardt, D. and B. Statzner. "Ecological impact of urban stormwater studied in experimental flumes: Population loss by drift and availability of refugial space." Aquatic Sciences. Vol. 52, No. 4. pp. 299-314. 1990.

Borchardt, D. and F. Sperling. "Urban stormwater discharges: Ecological effects on receiving waters and consequences for technical measures." 7^th International Conference on Urban Storm Drainage. Hannover, Germany. Sept. 9-13, 1996. Edited by F. Sieker and H-R. Verworn. IAHR/IAWQ. SuG-Verlagsgesellschaft. Hannover, Germany. pp. 359-364. 1996.

Boudries, H., C. Broguet, J-M. Mouchel, and D.R. Thévenot. "Urban runoff impact on composition and concentration of hydrocarbons in River Seine suspended solids." 7^th International Conference on Urban Storm Drainage. Hannover, Germany. Sept. 9-13, 1996. Edited by F. Sieker and H-R. Verworn. IAHR/IAWQ. SuG-Verlagsgesellschaft. Hannover, Germany. pp. 569-574. 1996.

Brelot, E., B. Chocat, and D. Villessot. "Methodological elements for analysing the impact of urban discharges on receiving waters." 7^th International Conference on Urban Storm Drainage. Hannover, Germany. Sept. 9-13, 1996. Edited by F. Sieker and H-R. Verworn. IAHR/IAWQ. SuG-Verlagsgesellschaft. Hannover, Germany. pp. 335-340. 1996.

Brookes, A. Channelized Rivers: Perspectives for Environmental Management. John Wiley and Sons. 1988.

Brookes, A. "Design issues." In: Effects of Urban Runoff on Receiving Systems: An Interdisciplinary Analysis of Impact, Monitoring, and Management. Engineering Foundation Conference. Mt. Crested Butte, CO. ASCE, NY. 1991.

Budhendra, B., M. Grove, C. Lowry, and J. Harbor. "Assessing long-term hydrologic effects on land use change." Journal of the American Water Works Association. Vol. 89, no. 11, pp. 94 – 106.

Burkholder, J.M., L.M. Larsen, H.B. Glasgow, Jr., K.M. Mason, P. Gama, and J.E. Parsons. "Influence of sediment and phosphorus loading on phytoplankton communities in an urban piedmont reservoir." Journal of Lake and Reservoir Management. Vol. 14, no. 1, pp. 110 – 121. January 1998.

Burton, G.A., Jr. and B.L. Stemmer. "Evaluation of surrogate tests in toxicant impact assessments." Toxicity Assessment: An International Journal. Vol. 3, pp. 255-269. 1988.

Burton, G.A., Jr. "Evaluation of seven sediment toxicity tests and their relationships to stream parameters." Toxicity Assess. 4:149-159. 1989.

Burton, G.A., Jr., B.L. Stemmer, and K. L. Winks. "A multitrophic level evaluation of sediment toxicity in Waukegan and Indiana Harbors," Environ. Toxicol. Chem. 8:1057-1066. 1989.

Burton, G.A., Jr. "Assessing freshwater sediment toxicity." Environ. Toxicol. Chem. 10(12).1991.

Burton, G.A., Jr., and J. Scott. "Sediment toxicity evaluations," Environ. Sci. Technol. Vol. 25. 1992.

Burton, G.A., Jr., Ed. Sediment Toxicity Assessment. Boca Raton, FL. Lewis Publishers. 1992.

Burton, G.A. and R.E. Pitt. A Manual for Conducting Evaluations of the Effects of Urban Runoff on Aquatic Life. CRC Press. To be published in 1998.

Cabelli, V.J., H. Kennedy, and M.A. Levin. "Pseudomonas aeruginosa-fecal coliform relationships in estuarine and fresh recreational waters." Journal WPCF 48, 2:367-376. Feb. 1976.

Cabelli, V.J., A.P. Dufour, M.A. Levin, L.J. McCabe, and P.W. Haberman. "Relationship of microbial indicators to health effects at marine bathing beaches." American Journal of Public Health. Vol. 69, no. 7, pp. 690-696. July 1979.

Cabelli, V.J., A.P. Dufour, L.J. McCabe, and M.A. Levin. "Swimming-associated gastroenteritis and water quality." American Journal of Epidemiology. Vol. 115, no. 4, pp. 606-616. 1982.

Cairns, J.,Jr. "A strategy for use of protazoans in the evaluation of hazardous substances." In: Biological Indicators of Water Quality. John Wiley and Sons. New York. 1979.

Calhoun, A. C., ed. Inland Fisheries Management, California Department of Fish and Game, Sacramento, California, 1966.

Callahan, M.A., M.W. Slimak, N.W. Gabel, I.P. May, C.F. Fowler, J.R. Freed, P. Jennings, R.L. Durfee, F.C. Whitmore, B. Maestri, W.R. Mabey, B.R. Holt, and C. Gould. Water Related Environmental Fates of 129 Priority Pollutants. U.S. Environmental Protection Agency, Monitoring and Data Support Division, EPA-4-79-029a and b. Washington D.C. 1979.

Cardozo, R.J., W.R. Adams, and E.L. Thackston. "CSO’s real impact on water quality: the Nashville experience." A Global Perspective for Reducing CSOs: Balancing Technologies, Costs, and Water Quality. July 10-13, 1994. Louisville, KY. Water Environment Federation. Alexandria, VA. 1994.

Carpenter, K.E. "On the biological factors involved in the destruction of river fisheries by pollution due to lead mining." Annuals of Applied Biology. Vol. 11. pp. 1-23. 1925.

Cave, K.A., "Receiving water quality indicators for judging stream improvement." In: Sustaining Urban Water Resources in the 21^st Century. Proceedings of an Engineering Foundation Conference. Edited by A.C. Rowney, P. Stahre, and L.A. Roesner. Malmo, Sweden. Sept. 7 – 12, 1997. To be published by ACSE, New York. 1998.

CDWR (California Department of Water Resources). Anderson Reservoir Limnologic Investigation, California Department of Water Resources, Central District, May 1978.

Cheung, W.H.S., K.C.K. Chang, R.P.S. Hung, and J.W.L. Kleevens. "Health effects of beach water pollution in Hong Kong." Epidemiol. Infect. Vol. 105, pp. 139-162. 1990.

Chisholm, J.L. and S.C. Downs. Stress and Recovery of Aquatic Organisms as Related to Highway Construction Along Trutle Creek, Boone County, West Virginia. USGS. Geological Survey Water Supply Paper 2055. Washington, D.C. 1978.

Choi, W-J. and C-K Park. "Effect of nonpoint pollution sources on water quality of streams in Pusan City, South Korea." Bulletin National Fisheries Univ. Busan (Nat. Sci.) (in Korean). Vol. 28. No. 2. pp. 15-22. 1988.

Claytor, R.A. and W. Brown. Environmental Indicators to Assess the Effectiveness of Municipal and Industrial Stormwater Control Programs. Prepared for the U.S. EPA, Office of Wastewater Management. Center for Watershed Protection, Silver Spring, MD. 210 pgs. 1996.

Claytor, R.A. "An introduction to stormwater indicators: an urban runoff assessment tool." Watershed Protection Techniques. Vol. 2, no. 2, pp. 321 - 328. Spring 1996a.

Claytor, R.A. "Multiple indicators used to evaluate degrading conditions in Milwaukee County." Watershed Protection Techniques. Vol. 2, no. 2, pp. 348 - 351. Spring 1996b.

Claytor, R.A. "Habitat and biological monitoring reveals headwater stream impairment n Delaware’s Piedmont." Watershed Protection Techniques. Vol. 2, no. 2, pp. 358 - 360. Spring 1996c.

Claytor, R.A. "An introduction to stormwater indicators: Urban runoff assessment tools." Presented at the Assessing the Cumulative Impacts of Watershed Development on Aquatic Ecosystems and Water Quality conference. March 20 – 21, 1996. Northeastern Illinois Planning Commission. pp. 217 – 224. Chicago, IL. April 1997.

Cook, W.L., F. Parrish, J.D. Satterfield, W.G. Nolan, and P.E. Gaffney. Biological and Chemical Assessment of Nonpoint Source Pollution in Georgia: Ridge-Valley and Sea Island Streams. Department of Biology, Georgia State University. Atlanta, Georgia. 1983.

Corbett, S.J., G.L. Rubin, G.K. Curry, D.G. Kleinbaum, and the Sydney Beach Users Study Advisory Group. "The health effects of swimming at Sydney beaches." American Journal of Public Health. Vol. 83, no. 12, pp. 1701 – 1706. December 1993.

Cordone, A.J. and D.W. Kelley. "Influences of inorganic sediments on aquatic life of streams." Californica Fish and Game. Vol. 47. pp. 189-228. 1961.

Correll, D.L., J.J. Miklas, A.H. Hines, and J.J. Schafer. "Chemical and biological trends associated with acidic atmospheric deposition in the Rhode River watershed and estuary, Maryland, USA." Water, Air, and Soil Pollution. Vol. 35. No. 1-2. pp. 63-86. 1987.

Craun, G.F., R.L. Calderon, and F.J. Frost. "An introduction to epidemiology." Journal of the AWWA. Vol. 88, no. 9. pp. 54-65. September 1996.

Craun, G.F., P.S. Berger, and R.L. Calderon. "Coliform bacteria and waterborne disease outbreaks." Journal of the AWWA. Vol. 89, no. 3. pp. 96-104. March 1997.

Crockett, C.S. and C.N. Haas. "Understanding protozoa in your watershed." Journal of the American Water Works Association. Vol. 89, No. 9, pp. 62 – 73. September 1997.

Crunkilton, R., J. Kleist, J. Ramcheck, W. DeVita, and D. Villeneueve. "Assessment of the response of aquatic organisms to long-term in-situ exposures to urban runoff." Presented at the Effects of Watershed Developments and Management on Aquatic Ecosystems conference. Snowbird, UT, August 4 – 9, 1996. Edited by L.A. Roesner. ASCE, New York. 1997.

CTA, Inc. Georgia Nonpoint Source Impact Assessment Study: Blue Ridge/Upland Georgia Cluster, Piedmont Cluster, and Gulf Coastal Plain Cluster. Georgia Envir, Protection Division, Dept. of Natural Resources, Atlanta, Georgia. 1983.

Cunningham, P.A. Nonpoint Source Impacts on Aquatic Life - Literature Review. U.S. Environmental Protection Agency. Office of Water Regulations and Standards. Contract No. 68-03-3423. Washington, D.C. July 1988.

D’Antuono, J.R. "Storm water permitting in the Milwaukee River basin." and "Performance comparison of highway BMPs." In: Watershed Management: Moving from Theory to Implementation. May 3 – 6, 1998. pp. 655 – 662. Water Environment Federation. 1998.

Dauer, D.M., W.W. Robinson, C.P. Seymour, and A.T. Leggett, Jr. Effects of Nonpoint Pollution on Benthic Invertebrates in the Lynnhaven River System. Virginia Water Research Center. Virginia Polytechnic Institute and State University. Blacksburg, Virginia. February 1979.

Davies, P.H. "Toxicology and chemistry of metals in urban runoff." In: Urban Runoff Quality: Impact and Quality Enhancement Technology. Engineering Foundation Conference. Henniker, NH. ASCE, NY. 1986.

Davies, P.H. "Synergistic effects of contaminants in urban runoff." In: Effects of Urban Runoff on Receiving Systems: An Interdisciplinary Analysis of Impact, Monitoring, and Management. Engineering Foundation Conference. Mt. Crested Butte, CO. ASCE, NY. 1991.

Davies, P.H. "Factors in controlling nonpoint source impacts." Engineering Foundation/ASCE conference held in 1991 in Mt. Crested Butte, Colorado: Stormwater Runoff and Receiving Systems: Impact, Monitoring and Assessment (edited by E.E. Herricks). Lewis/CRC Press. Boca Raton. pp. 53 – 64. 1995.

Delzer, G.C., J.S. Zogorski, T.J. Lopes and R.L. Bosshart. Occurrence of the Gasoline Oxygenate MTBE and BTEX Compounds in Urban Stormwater in the United States, 1991 - 95. USGS Water-Resources Investigations Report 96 - 4145. U.S. Geological survey, Rapid City, SD. 1996.

DePinto, J.V., T.C. Young and S.C. Martin. "Aquatic sediments." Journal of Water Pollution Control Federation. Vol. 52. No. 6. pp 1656-70. June 1980.

Desa, M and J. Niemczynowicz. "Spatial variability of rainfall in Kuala Lumpur, Malaysia: Long and short term characteristics." Hydrological Sciences Journal. Vol. 41, no. 3. Pp. 345 - 362. June 1996.

Desbordes, M. and J.C. Hemain. "Further research needs for impact estimates of urban storm water pollution." Water Science and Technology. Vol. 22, No. 10/11. 1990.

DHS (Department of Health Services). Guidance for Freshwater Recreational Areas: Assessing Microbiological Contamination and taking Corrective Action (Draft). State of California Health and Welfare Agency. Sacramento, CA. November 1997.

DHS (Department of Health Services). Guidance for Saltwater Recreational Areas (Oceans, Bays, Estuaries, and the Salton Sea): Assessing Microbiological Contamination and taking Corrective Action (Draft). State of California Health and Welfare Agency. Sacramento, CA. November 1997.

D’Itri, F.M. "The biomethylation and cycling of selected metals and metalloids in aquatic sediments." In: Sediments: Chemistry and Toxicity of In-Place Pollutants. Lewis Publishers. Ann Arbor, Michigan, 1990.

Dowling, S.J. and B.W. Mar. "Culvert composite sampler: A cost-effective stormwater monitoring device." Journal of Water Resources Planning and Management - ASCE. Vol. 122, no. 4, pp. 280 - 286. July/Aug. 1996.

Dreher, D.W. "Watershed urbanization impacts on stream quality indicators in northeastern Illinois." Presented at the Assessing the Cumulative Impacts of Watershed Development on Aquatic Ecosystems and Water Quality conference. March 20 – 21, 1996. Northeastern Illinois Planning Commission. pp. 129 – 135. Chicago, IL. 1997.

Driscoll, S.K., and P.F. Landrum. "A comparison of equilibrium partitioning and critical body residue approaches for predicting toxicity of sediment-associated fluoranthene to freshwater amphipods." Environmental Toxicology and Chemistry. Vol. 16, no. 10, pp. 2179 – 2186. 1997.

Droste and Gupgupoglu. Indicator Bacteria Die-Off in the Rideau River. Rideau River Stormwater Management Study, Ottawa, and the Ontario Ministry of the Environment, Kingston, Ontario. Feb. 1982.

Duda, A.M., D.R. Lenat, and D. Penrose. "Water quality degradation in urban streams in the Southeast: Will non-point source controls make any difference?" In: Proceedings of the International Symposium on Urban Storm Runoff. University of Kentucky. Lexington, KY, July 1979.