Lagrangian Sampling for Emerging Contaminants Through an Urban Stream Corridor in Colorado1

Authors

  • Juliane B. Brown,

    1. Respectively, Hydrologist and Research Hydrologist (Brown and Battaglin), U.S. Geological Survey, Colorado Water Science Center, Denver Federal Center, MS 415, Lakewood, Colorado 80225
    Search for more papers by this author
  • William A. Battaglin,

    1. Respectively, Hydrologist and Research Hydrologist (Brown and Battaglin), U.S. Geological Survey, Colorado Water Science Center, Denver Federal Center, MS 415, Lakewood, Colorado 80225
    Search for more papers by this author
  • Robert E. Zuellig

    1. Aquatic Ecologist (Zuellig), U.S. Geological Survey, Fort Collins Science Center, 2150 Centre Avenue, Building C, Fort Collins, Colorado 80526.
    Search for more papers by this author

  • 1

    Paper No. JAWRA-07-0182-P of the Journal of the American Water Resources Association (JAWRA). Discussions are open until August 1, 2009.

(E-Mail/Brown: jbbrown@usgs.gov)

Abstract

Abstract:  Recent national concerns regarding the environmental occurrence of emerging contaminants (ECs) have catalyzed a series of recent studies. Many ECs are released into the environment through discharges from wastewater treatment plants (WWTPs) and other sources. In 2005, the U.S. Geological Survey and the City of Longmont initiated an investigation of selected ECs in a 13.8-km reach of St. Vrain Creek, Colorado. Seven sites were sampled for ECs following a Lagrangian design; sites were located upstream, downstream, and in the outfall of the Longmont WWTP, and at the mouths of two tributaries, Left Hand Creek and Boulder Creek (which is influenced by multiple WWTP outfalls). Samples for 61 ECs in 16 chemical use categories were analyzed and 36 were detected in one or more samples. Of these, 16 have known or suspected endocrine-disrupting potential. At and downstream from the WWTP outfall, detergent metabolites, fire retardants, and steroids were detected at the highest concentrations, which commonly exceeded 1 μg/l in 2005 and 2 μg/l in 2006. Most individual ECs were measured at concentrations less than 2 μg/l. The results indicate that outfalls from WWTPs are the largest but may not be the sole source of ECs in St. Vrain Creek. In 2005, high discharge was associated with fewer EC detections, lower total EC concentrations, and smaller EC loads in St. Vrain Creek and its tributaries as compared with 2006. EC behavior differed by individual compound, and some differences between sites could be attributed to analytical variability or to other factors such as physical or chemical characteristics or distance from contributing sources. Loads of some ECs, such as diethoxynonylphenol, accumulated or attenuated depending on location, discharge, and distance downstream from the WWTP, whereas others, such as bisphenol A, were largely conservative. The extent to which ECs in St. Vrain Creek affect native fish species and macroinvertebrate communities is unknown, but recent studies have shown that fish respond to very low concentrations of ECs, and further study on the fate and transport of these contaminants in the aquatic environment is warranted.

Introduction

During the last 20 years, land use in the Great Plains portion of the St. Vrain Creek basin, in north-central Colorado (Figure 1), has shifted substantially from largely undeveloped native grassland, pasture, and agricultural land to urban and suburban land, particularly in and around the cities of Longmont and Boulder, Colorado. The increase in human population associated with this shift has increased volumes of treated wastewater effluent that is discharged to area streams. Additionally, between 1986 and 1998 impervious surfaces increased 32% in the region, whereas irrigated cropland decreased by 33% and wetlands decreased by 65% (American Forests, 2001). Typically, changes in land cover can lead to increased urban runoff, decreased natural attenuation of discharge and filtering by wetlands, alterations in discharge conditions and water quality (Sprague et al., 2006), and ultimately degraded biological communities (Paul and Meyer, 2001).

Figure 1.

 St. Vrain Creek Basin, Colorado, With Land Use and Sampling Stations.

During the last two decades new wastewater treatment plant (WWTP) technologies and instream restoration efforts have been implemented by the city of Longmont to improve aquatic habitat and water quality in St. Vrain Creek (Zuellig et al., 2007). The WWTP improvements primarily addressed facility capacity and the reduction of elevated metals, nutrients, and suspended-sediment concentrations. Recent regional and national reconnaissance studies have indicated the presence of previously undocumented contaminants indicative of human sources downstream from WWTPs (Ternes, 2001; Kolpin et al., 2002; Heberer and Adam, 2005; Sprague and Battaglin, 2005). These wastewater-related contaminants include antioxidants, detergents and detergent metabolites, disinfectants, fire retardants, fragrances, insect repellants, pharmaceuticals (prescription and nonprescription drugs), pesticides, plasticizers, polycyclic aromatic hydrocarbons, and steroidal compounds and are referred to hereafter as “emerging contaminants” (ECs). ECs can be released to the aquatic environment through industrial and municipal wastewater discharges, storm drains, agricultural and urban runoff, and individual or multi-facility sewage disposal systems. Until recently, the extent to which these contaminants occurred in the aquatic environment was not well known and the toxicological ramifications with regard to humans or wildlife were largely unknown (Daughton, 2001). However, recent studies have shown that exposure to some ECs, even at very low concentrations, can result in endocrine disruption and histological and immunological alterations in wildlife and humans (Colborn et al., 1993; McLachlan, 2001; Petrovic et al., 2002; Höger, 2003; Bernet et al., 2004; Hoeger et al., 2004; Arslan et al., 2007). Identifying the occurrence, distribution, and fate of ECs in urban streams will aid communities in addressing source-control and reduction efforts to safeguard human and aquatic health.

In this paper we describe the occurrence and transport of selected ECs in St. Vrain Creek through the city of Longmont under two different hydrologic conditions. Stream samples were collected during the two events by using a longitudinal Lagrangian sampling design. Dye tracer studies conducted just prior to each sampling event were used to estimate travel times during each event. Discharge and field measurements were collected and water samples were analyzed by the U.S. Geological Survey (USGS) National Water Quality Laboratory (NWQL) for a series of ECs that are indicative of wastewater by using methods described in Zaugg et al. (2002).

Study Area and Site Selection

St. Vrain Creek flows east from sources along the east side of the Continental Divide and eventually joins the South Platte River, in north-central Colorado (Figure 1). The main mountainous headwater streams, North and South St. Vrain Creeks, are primarily forested until they converge downstream near the town of Lyons to form St. Vrain Creek. Downstream from Lyons, St. Vrain Creek primarily flows through grassland, pastures, and agricultural areas and the city of Longmont on its way to the confluence with the South Platte River.

Seven sites were selected for water-chemistry sampling along a 13.8-kilometer (km) reach of the St. Vrain Creek within the city of Longmont: two sites upstream from the Longmont WWTP outfall (Sites 1 and 2); two sites at the mouths of key tributaries, Left Hand Creek and Boulder Creek (Sites 3 and 6); one site at the Longmont WWTP outfall (Site 4); and two sites on St. Vrain Creek downstream from the WWTP outfall (Sites 5 and 7) (Figure 1). The most upstream site (Site 1) is located at the western edge of Longmont just east (downstream) from Airport Road, and it represents the upstream, primarily nonurban inputs to the creek before it enters the Longmont area. This site is influenced by the small town of Lyons approximately 13.2 km upstream and the surrounding agricultural community. Site 2 is approximately 5.8 km downstream from Site 1 and immediately upstream from the Longmont WWTP outfall; it represents inputs from the urban and agricultural areas before the input of treated wastewater effluent. Left Hand Creek (Site 3) is one of the two largest tributaries entering St. Vrain Creek in the study reach. Left Hand Creek enters St. Vrain Creek just upstream from the Longmont WWTP outfall; it is influenced by the upstream communities of Jamestown and Ward, some historical mining, and suburban development and agricultural activities around Longmont. Site 4 is the WWTP outfall and the flow is composed of treated wastewater effluent from the City of Longmont Water and Wastewater Department. Site 5 is approximately 1 km downstream from the Longmont WWTP outfall and represents the integration of the WWTP effluent and St. Vrain Creek (tributaries do not intervene between the outfall and this site). Boulder Creek (Site 6) is the other major tributary and it enters St. Vrain Creek approximately 7.5 km downstream from the Longmont WWTP outfall and is influenced by upstream operations of Barker Dam, the town of Nederland, historical mining, and WWTP inflows from the urban areas of Boulder, Superior, Louisville, Lafayette, and Erie, and by the surrounding suburban and agricultural land. The WWTP for Boulder is approximately 23 km upstream from Site 6. Site 7 is just downstream from the confluence of Boulder Creek, approximately 7.9 km downstream from the Longmont WWTP and approximately 72.6 km from the confluence with the South Platte River.

Data Collection

A Lagrangian sampling design, which follows the same parcel of water as it moves downstream, was used for each sampling event (Zuellig et al., 2007). Tracer tests with Rhodamine-WT dye were used to determine the time-of-travel between sample-collection sites (Kilpatrick and Wilson, 1989). For this study, time-of-travel was defined as “the amount of elapsed time for the dye peak to travel between two monitoring sites” (Zuellig et al., 2007). Travel-time estimates were made 9 and 15 days prior to the date of sample collection in April 2005 and March 2006, respectively. Minor adjustments were made to the dye travel-time estimates to account for differences in flow conditions during the times of sample collection (Zuellig et al., 2007). In 2005, under higher flow conditions than in 2006, times of travel were estimated as 345 min between Sites 1 and 2, 65 min between Sites 2 and 5, and 235 min between Sites 5 and 7. In 2006, times of travel were estimated as 825 min between Sites 1 and 2, 45 min between Sites 2 and 5, and 290 min between Sites 5 and 7. Stream velocity and travel times can increase or decrease during lower flow conditions depending on stream channel morphology. Uncertainty in the estimates of travel time can be introduced by errors in the measurement of dye concentrations.

Stream samples analyzed for ECs were collected at specific times calculated from the travel-time data using standardized depth- and width-integrating techniques and processed and preserved on-site using methods described in the USGS National Field Manual (variously dated). Samples were analyzed by the USGS NWQL for 61 ECs (Table 1) according to methods described in Zaugg et al. (2002). Field measurements, including dissolved oxygen, pH, specific conductance, water temperature, and discharge, were obtained at the time of sample collection. The complete dataset is presented in Zuellig et al. (2007). Most of the 61 ECs are commercially synthesized compounds or their degradation products, but a few such as phenol, skatol, and the steroids can originate from natural sources.

Table 1.   Emerging Contaminants Analyzed in Water Samples From St. Vrain Creek and in Left Hand Creek and Boulder Creek Tributaries, Longmont, Colorado, April 2005, and March 2006. Thumbnail image of Thumbnail image of Thumbnail image of

The analytical results consisted of unqualified concentrations, E coded (or estimated) concentrations, and nondetections [reported as less than the laboratory reporting level (LRL) for the particular compound]. Estimated concentrations include those that are below or above the calibration curve, concentrations for compounds with average recoveries that are less than 60%, or concentrations of compounds routinely detected in laboratory blanks (Furlong et al., 2001). Both unqualified concentrations and E coded concentrations are used in the calculation of summary statistics (Table 1) and contaminant loads (Tables 2 and 3).

Table 2.   Summary of Number of Detections, Discharge, Total Concentration, and Total Load of Detected Wastewater-Related Contaminants by Sample Event, St. Vrain Creek and Tributaries Left Hand Creek and Boulder Creek, Near Longmont, Colorado, April 2005 and March 2006.
Site No.20052006
NOD1 Discharge (m3/s)Total Concentration1 (μg/l)Total Load1,2 (g/day) NOD1 Discharge (m3/s)Total Concentration1 (μg/l)Total Load1,2 (g/day)
  1. Notes: NOD, number of detections; WWTP, wastewater treatment plant.

  2. 1Excludes phenol and naphthalene from both years.

  3. 2Loads computed for detected compounds only (excludes all nondetected data, includes estimated data).

101.200020.120.2192.27
211.550.067.890.252.8461.1
3 (Tributary)00.130010.120.030.3
4 (WWTP)300.3829.3967310.3541.41,255
5202.126.121,123310.7623.41,528
6 (Tributary)192.466.541,392231.2913.71,528
7214.394.721,791262.0813.52,414
Table 3.   Mass Balance Summary of Detected Emerging Contaminants by Sample Event, St. Vrain Creek and Tributaries Left Hand Creek and Boulder Creek, Near Longmont, Colorado, April 2005, and March 2006.
Compound Name (common name)2005 Accumulated Load by Sites (g/day)2006 Accumulated Load by Sites (g/day)Log Kow (LogP)
Sites 2 + 3 + 4Site 5% DifferenceSites 5 + 6Site 7% DifferenceSites 2 + 3 + 4Site 5% DifferenceSites 5 + 6Site 7% Difference
Accumulated Discharge (m3/s)2.062.124.584.390.720.762.052.08
  1. Kow, octanol-water partitioning coefficient; -, not analyzed or not computed; <, result was below laboratory reporting level; ↔, no change; ↑, increase; ↓, decrease.

  2. Greater than 20% difference indicates probable accumulation () or attenuation (↓).

Antioxidant
 3-tert-Butyl-4-hydroxyanisole (BHA)3.30<-↓<<-3.614.8534.34.85<-↓3.5
 5-Methyl-1H-benzotriazole49.512515312524394.446.477.166.2169128−24.31.71
 p-Cresol46.247.73.25↑69.045.5−34.130.315.8−47.933.139.318.7↑1.94
Cosmetic
 Triethyl citrate (ethyl citrate)11.914.521.828.826.2−9.03↓20.419.6−3.92↓33.624.8−26.20.33
Deodorizer
 1,4-Dichlorobenzene5.625.51−1.96↓5.51<-↓6.286.361.27↑6.362.06−67.63.44
Detergent metabolite
 4-Nonylphenol (total)36.375.210716918610.1↑68.911972.7119116−2.52↓5.92
 4-tert-Octylphenol<<-<<-5.706.7318.1↑16.513.1−20.65.28
 Diethoxynonylphenol (total, NPEO2)23829423.5656<-↓40244510.7↑767481−37.3-
 Diethoxyoctylphenol (OPEO2)10.2<-↓38.337.9−1.04↓25.628.912.9↑57.143.6−23.6-
 Ethoxyoctylphenol (OPEO1)<<-49.060.723.99.9214.243.152.239.1−25.1-
Disinfectant
 Triclosan18.518.4−0.54↓50.345.5−9.5↓23.430.831.652.338.4−26.64.76
Fire retardant
 Tris(2-butoxyethyl) phosphate149147−1.34↓326288−11.7↓29033916.9↑834705−15.5↓3.75
 Tris(2-chloroethyl) phosphate9.5812.530.535.937.95.57↑8.3712.043.429.924.4−18.4↓1.44
 Tributyl phosphate10.615.041.523.421.2−9.40↓5.528.1748.016.315.0−7.98↓4.00
 Tris(dichloroisopropyl) phosphate15.223.957.239.837.2−6.53↓10.616.151.939.233.7−14.0↓3.65
Fixative
 Benzophenone6.616.610.0↔6.6121.22215.866.6012.6↑15.114.7−2.65↓3.18
Flavorant
 Camphor1.19<-↓<<-1.742.6250.62.623.5937.02.38
Fragrance
 3-Methyl-1H-indole (skatol)1.92<-↓<10.6-↑0.8100.8363.21↑0.836<-↓2.60
 Acetophenone3.96<-↓<<-6.407.6419.4↑7.64<-↓1.58
 Acetylhexamethyltetrahydro-   naphthalene (AHTN)11.912.11.68↑20.617.4−15.5↓13.615.916.9↑15.912.9−18.9↓5.70
 Hexahydrohexamethyl-   cyclopentabenzopyran (HHCB)69.478.913.7↑130106−18.5↓77.493.420.7↑13283.6−36.75.90
 Indole7.606.61−13.0↓6.61<-↓3.283.6411.0↑7.66<-↓2.14
Nonprescription drug
 Caffeine32.236.714.0↑75.064.5−14.0↓19.221.09.38↑57.250.2−12.2↓−0.07
 Cotinine10.9<-↓<<-4.165.2526.25.25<-↓0.07
Pesticide
 Metalaxyl2.18<-↓<<-<<-<<-1.65
 N,N-diethyl-meta-toluamide (DEET)12.222.080.345.441.7−8.15↓10.312.925.228.026.9−3.93↓2.18
Plasticizer
 Bisphenol A4.29<-↓<<-20.921.94.8↑50.650.4−0.40↓3.32
 Triphenyl phosphate 4.957.5251.913.511.0−18.5↓5.868.1038.213.410.0−25.44.59
Polycyclic aromatic hydrocarbons (PAH)
 Fluoranthene<<-<<-0.185<-↓<<-5.16
 Phenanthrene<--<<-0.1980.8363220.836<-↓4.46
Steroid
 3-β-Coprostanol33.060.683.6154106−31.244.852.316.7↑125124−0.80↓8.82
 β-Sitosterol42.9<-↓140<-↓31.633.86.96↑10181.4−19.4↓9.65
 β-Stigmastanol52.9<-↓<110--↑27.9<-↓67.565.3−3.26↓-
 Cholesterol72.711456.8308273−11.4↓86.597.312.5↑196188−4.08↓8.74

Quality Control and Quality Assessment

Quality-control samples were collected as part of this study, including one field blank, one replicate, and one laboratory spike during each sampling event. In 2005, phenol was detected in the field blank and in 2006 methyl salicylate and naphthalene were detected in the field blank. All phenol results for 2005 and naphthalene results for 2006 were considered contaminated. Results for methyl salicylate analyses were qualified as acceptable because all results were nondetectable and the estimated value in the blank (0.0170 μg/l) was much smaller than the LRL (<0.5 μg/l). To facilitate between-year comparisons, no phenol or naphthalene results were further analyzed as part of this study, though the summary data are included in Table 1 for reference.

In 2005, percent difference between environmental and replicate samples ranged from 0 to 57.5% with an overall median percent difference of 9.1 for the 31 detected ECs, excluding phenol (Table 1). In 2006, percent difference between environmental and replicate samples ranged from 6.2 to 30.4% with an overall median percent difference of 15.8 for the three detected ECs, excluding naphthalene (Table 1).

Field and laboratory spike recovery data for the ECs added to environmental sample water and laboratory reagent water at known concentrations are shown in Table 1. Percent recoveries are determined by dividing the measured concentration in the environmental or laboratory sample by the known concentration in the added spike solution. Greater than 60% of most constituents (57 of 61) were recovered for the field or laboratory spikes for one or both sampling events. Only four constituents, β-stigmastanol, d-limonene, isopropylbenzene, and tetrachloroethylene, had environmental and laboratory spike recoveries less than 60% for one or both sampling events, indicating generally poor recovery and increased uncertainty in quantification. Of these four constituents, only β-stigmastanol was detected in an environmental sample. For the 2005 event, field spike recoveries ranged from 44.1 to 171.9% and laboratory recoveries ranged from 19.2 to 102.2% with overall mean and median recoveries of 75.3 and 82.7%, respectively. For the 2006 event, field spike recoveries ranged from −22.5 to 206.1% and the laboratory recoveries ranged from 7.6 to 102.3% with overall mean and median recoveries of 72.2 and 78.3%, respectively. The relatively wide range of recoveries is not atypical for analysis of these types of compounds (Lee et al., 2004).

Differences between paired environmental and replicate samples and field and laboratory spike recoveries could be attributed to analytical variability, contamination in the environmental sample, contamination of the spike solution, or variability in recoveries owing to differences in physical or chemical properties of the ECs (Zaugg and Leiker, 2006).

Results

Of the 61 ECs analyzed for, 36 were detected in one or more samples (Table 1); two of these ECs (naphthalene and phenol) were excluded from further analysis due to blank contamination in one of the two years of sampling. Twenty-five ECs were not detected at any of the sites during either sampling event. The number of detections was slightly higher for the 2006 event, which was sampled at a lower flow than the 2005 event (Table 2, Figure 2). Measured discharge from the WWTP outfall and Left Hand Creek was not substantially different between sampling events; however, the measured discharges in 2005 on St. Vrain Creek and Boulder Creek were two to ten times as high as in 2006. Concentrations of detected ECs were generally low in St. Vrain Creek and Left Hand Creek upstream from the WWTP outfall, highest in the WWTP outfall, and generally decreased downstream from the WWTP outfall, although an increase in concentrations at Site 7 (downstream from the Boulder Creek inflow) was observed for some constituents.

Figure 2.

 Number of Emerging Contaminants Detected, by Chemical Use Category, During (A) 2005 and (B) 2006 Sampling Events, St. Vrain Creek and Tributaries Left Hand Creek and Boulder Creek, Longmont, Colorado (see Figure 1 for site locations).

Patterns of detection and concentrations of ECs and chemical use categories upstream and downstream from the Longmont WWTP, at the outfall, and from the two sampled tributaries varied considerably by location, year, and discharge (Figures 2 and 3). For example, in 2005 only one nonprescription drug, caffeine, was detected upstream from the WWTP outfall, whereas in 2006, 12 ECs in nine categories were detected upstream from the WWTP outfall. At the WWTP outfall (Site 4), 30 ECs in 13 categories were detected in 2005, and 31 ECs in 13 categories were detected in 2006. All but three of the ECs in the WWTP outfall (ethoxyoctylphenol, metalaxyl, and 4-tert-octylphenol) were found in both years. Downstream from the WWTP outfall at Site 5, 20 ECs in 12 categories were detected in 2005, and 31 ECs in 14 categories were detected in 2006. Further downstream at Site 6 (Boulder Creek), 19 ECs in 10 categories were detected in 2005, and 23 ECs in 11 categories were detected in 2006. At Site 7 (the furthest downstream), 21 ECs in 11 categories were detected in 2005, and 26 ECs in 13 categories were detected in 2006.

Figure 3.

 Concentration of Emerging Contaminants, by Chemical Use Category, and Discharge Detected During (A) 2005 and (B) 2006 Sampling Events, St. Vrain Creek and Tributaries Left Hand Creek and Boulder Creek, Longmont, Colorado (see Figure 1 for site locations).

At the WWTP outfall (Site 4) detergent metabolites, fire retardants, steroids, fragrances, and antioxidants were the five most frequently detected EC categories in both years (Figure 2). Of the 36 ECs detected during one or both sample events, 16 have been identified as having known or suspected endocrine-disrupting potential (Table 1). In both years, detergent metabolites, fire retardants, and steroids were detected at the highest concentrations at Sites 4, 5, 6, and 7 (Figure 3). Total concentrations for these three categories at these four sites frequently exceeded 1 μg/l in 2005 and always exceeded 2 μg/l in 2006. Most individual ECs were measured at concentrations less than 2 μg/l; the exceptions were diethoxynonylphenol (12.4 μg/l), tris(2-butoxyethyl) phosphate (9.37 μg/l), hexahydrohexamethyl-cyclopentabenzopyran (2.54 μg/l), cholesterol (2.45 μg/l), and 4-nonylphenol (2.27 μg/l) (Table 1; Figure 3). Sixty-five to 71% of the analytical results from the two sampling events were reported to be below the LRL.

Estimates of total measured EC concentrations (and load) are not specifically relevant to toxicity but may be associated with or related to potential for ecosystem effects. These quantities also are useful for site-to-site and year-to-year comparisons. Total measured concentrations of ECs were greater for the 2006 sampling event than for the 2005 event at all sites (Table 2). During the 2006 event, there was a larger relative reduction (42%) in total concentration between Site 5, the first site downstream from the WWTP outfall, and Site 7, the most downstream site, than the reduction that occurred in 2005 (23%). The differences in relative reductions between 2005 and 2006 could be related to the frequencies of constituent detection. In 2005, between Sites 5 and 7, there was an increase in the number of constituents detected from 20 to 21, whereas in 2006 there was a decrease from 31 to 26 (Table 2).

Whereas total concentrations decreased below the WWTP outfall as ECs traveled downstream via dilution and other mechanisms, total loads increased downstream with increasing discharge (Figures 4 and 5; Table 2). During both the 2005 and 2006 sampling events, discharge was conservative; flows differed by less than 5% between Site 5 and the sum of Sites 2, 3, and 4 and between Site 7 and the sum of Sites 5 and 6. In 2005, the WWTP discharge was equal to 18% of the flow at Site 5, whereas in 2006 it equaled 46% of the flow at Site 5. In 2005, the outflow from the Longmont WWTP comprised 86% of the total measured EC load at Site 5, and 54% at Site 7. In 2006, the outflow from the Longmont WWTP contained 82% of the total EC load at Site 5, and 52% at Site 7 (Table 2). In 2005 and 2006, the total EC loads from Boulder Creek (Site 6) were larger than the loads contributed by the Longmont WWTP (Figure 4; Table 2), and total EC load was not transported conservatively. In both years, total EC load increased between 13 and 14% between (the sum of loads at) Sites 2, 3, and 4 and at Site 5. Total EC load then decreased between (the sum of loads at) Sites 5 and 6 and Site 7 by 29% in 2005 and by 21% in 2006. Total EC load was higher in 2006 (the lower flow year) than in 2005 (Figure 4; Table 2).

Figure 4.

 Discharge (A and C) and Total Measured Contaminants Load (B and D) During 2005 and 2006 Sampling Events, St. Vrain Creek and Tributaries Left Hand Creek and Boulder Creek Near Longmont, Colorado.

Figure 5.

 Loads of (A) Bisphenol A, (B) Caffeine, and (C) Diethoxynonylphenol, During 2005 or 2006 Sampling Events, St. Vrain Creek and Tributaries Left Hand Creek and Boulder Creek Near Longmont, Colorado.

The accuracy of the load estimates could be affected by error in the discharge measurements and uncertainty associated with the analytical results. Most discharge measurements used for this study have a quality rating of “fair to good” and should be in error by no more than 5-8% (Rantz et al., 1982). The exceptions are discharge measurements for Site 3, which are rated “poor” and may be in error by more than 8%, and discharge measurements for Site 7, which are rated “good” with errors of 2-5%. The limited quality control data collected by this study indicate that analytical variability may introduce uncertainty in the measured concentrations that are larger than 10% and which could consequently affect the load calculations.

Certain individual ECs exhibited largely conservative behavior, whereas others accumulated or attenuated downstream from the WWTP outfall (Table 3). For example, in 2006, the Longmont WWTP contributed 72% of the load of bisphenol A at Site 5 (Figure 5), and the load at Site 5 was approximately the same as (4.8% greater) the sum of the loads at Sites 2, 3, and 4. Similarly conservative behavior was recorded downstream where the load at Site 7 was approximately the same as (0.4% less) the sum of the loads at Sites 5 and 6. Some ECs behaved differently at different locations. For example, in 2005, the Longmont WWTP contributed 66% of the load of caffeine at Site 5, where the load was greater (14%) than the sum of the loads from Sites 2, 3, and 4; but farther downstream the load at Site 7 was less (14%) than the sum of the loads at Sites 5 and 6 (Figure 5). Selected ECs showed substantial attenuation as they traveled downstream from the WWTP outfall. For example, in 2006, the Longmont WWTP contributed 84% of the load of diethoxynonylphenol at Site 5, where the load was greater (10.7%) than the sum of the loads from Sites 2, 3, and 4; but the load at Site 7 was much less (37.3%) than the sum of the loads at Sites 5 and 6 (Figure 5; Table 3).

Discussion

The results indicate that outfalls from WWTPs are the largest, but are likely not the sole, sources of ECs to St. Vrain Creek. Additional potential sources of ECs not identified during this study could include ground water, seepage from bank storage and individual sewage disposal systems, storm drains, nonpoint runoff, and atmospheric deposition. During the 2005 and 2006 sampling events, discharge was conservative and streamflows differed by less than 5% between Site 5 and the sum of Sites 2, 3, and 4 and between Site 7 and the sum of Sites 5 and 6. Therefore, relative changes in EC loads are most likely related to factors other than differences in streamflow. In 2005 and 2006, the total EC loads increased along the Longmont urban corridor by approximately 13-14%. The increase in loads between the sum of Sites 2, 3, and 4 (the WWTP outfall) and Site 5 (downstream of the WWTP outfall) may indicate contributions of ECs either from upstream of the WWTP or from the WWTP directly that were too low to be detected individually. ECs might also be contributed to the stream between the outfall and Site 5 (a distance of about 1 km), though no specific sources were identified.

Further downstream, the concentrations and loads of most measured ECs attenuated. For example, in 2006 the percent difference between the sum of loads from Sites 5 and 6 was greater than the Site 7 load for 30 compounds (10 of which differed by more than 20%) and less than the Site 7 load for two compounds (one of which differed by more than 20%); the other 27 ECs were not detected at all three sites and naphthalene and phenol were excluded. In 2005, the percent difference between the sum of loads from Sites 5 and 6 was greater than the Site 7 load for 18 compounds (two of which differed by more than 20%) and less than the Site 7 load for seven compounds (three of which differed by more than 20%); the other 34 ECs were not detected at all three sites and naphthalene and phenol were excluded (Table 3). The amount of attenuation (percent difference between the sum of loads from Sites 5 and 6 and the Site 7 load) was greater in 2005 (29%) than in 2006 (21%), even though 2006 had lower flow conditions and correspondingly slower travel times that would promote degradation, volatilization, biological uptake, or adsorption. This apparent decrease in attenuation in 2006 may be the result of reduced dilution by the stream resulting in more frequent detection of ECs downstream from the WWTP outfall. Total concentration and total load of ECs were higher at Site 7 in 2006 than in 2005, even though discharge was 53% less. Physical and chemical characteristics such as stream temperature, that could affect how contaminants behave as they travel downstream, varied by compound. Additionally, partitioning to sediment may result in attenuation of selected ECs. ECs, such as the steroids, have relatively high octanol-water partitioning coefficients (Kow) and would be expected to adsorb to sediments, whereas others, like the nonprescription drugs caffeine and cotinine, have very low Kow values and would be expected to remain in solution (Table 3). However, a consistent pattern between Kow values and attenuation or accumulation of ECs was not found in this study. Analytical variability may have influenced the observed differences in loads, as some contaminants that were detected at relatively high concentrations such as β-stigmastanol have relatively poor analytical reproducibility (Table 1).

Substantial attenuation of many wastewater-derived contaminants in a large effluent-dominated river was noted by Fono et al. (2006), but in that system, travel times were measured in days rather than minutes. In the studied portion of St. Vrain Creek, there was evidence of various levels of attenuation for most ECs (Figures 4 and 5; Table 3). In St. Vrain Creek, the amount of attenuation observed was not sufficient to prevent aquatic biota from being exposed continuously to a wide range of ECs downstream from the Longmont WWTP and in Boulder Creek near the confluence with St. Vrain Creek (Table 3). St. Vrain Creek harbors native fish species and macroinvertebrate communities that have declined or are absent in other parts of the South Platte River Basin (Zuellig et al., 2007). The extent to which ECs in St. Vrain Creek affects these organisms is unknown, but recent studies have shown that fish can respond to very low concentrations of some organic or estrogenic contaminants (Brian et al., 2007; Quiros et al., 2007).

Acknowledgments

The authors acknowledge those from the USGS and from the City of Longmont who helped with field work. Additionally, comments provided by Jennifer L. Flynn, Betty Palcsak, and Allan D. Druliner (USGS Colorado Water Science Center), Lori Sprague (USGS National Water Quality Assessment Program), and Calvin Youngberg (City of Longmont Water and Wastewater Department) were greatly appreciated.

Ancillary